Fall Detection among Elderly Person using FallCNN and Transfer Learning Models

Fall Detection among Elderly Person using FallCNN and Transfer Learning Models | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Fall Detection among Elderly Person using FallCNN and Transfer Learning Models M. Vergin Raja Sarobin, Jishnuraj k, L. Jani Anbarasi, P Rukmani, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-1884093/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract As per the data provided by the World Health Organization (WHO), falls are one of the major reasons for unintentional deaths or injuries in elderly people. Even though there are a lot of fall detection methods and algorithms exist, there is no efficient artificial intelligence strategy for detecting falls. Various literature states that Fall Detection among Elderly Person (FDEP) provides the possibility of bringing up an efficient and cost-effective way to tackle this problem. This paper generated a signal-based image dataset, SimgFall from the existing accelerometer or gyroscope-based sensor data of the SiSFall dataset for early detection of fall to fasten the medical assistance process. This SimgFall data is proposed by the proposed FallCNN model, a novel deep Convolutional Neural Network (CNN) based architecture that includes multiple folds of CNN network. These models utilize depth-wise convolution with varying dilation rates for efficiently extracting diversified features from the SimgFall dataset. 1992 signal-based images of which 498 are the samples collected for fall, jump, stumble and walk of four classes respectively. Further, performance evaluation on the generated dataset using different pre-trained and custom models has been analysed based on the loss and accuracy curve. The experimental results show that the highest classification accuracy (98%) is achieved using the proposed customized FallCNN model. CNN Transfer learning model Fall detection Deep learning Performance analysis Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Figure 11 1. Introduction In today's world, one of the most common causes of death is fall which is more common among elderly people due to their physical weakness. Research done by World Health Organization (WHO) suggests that 42% of people above the age of 65 suffer from at least one fall. So as age increases elderly fall rates also increase. Although, there are a lot of fall detection devices available in the market most of them are expensive, non-portable, or inaccurate. The fear of falling is more among people who live alone because it takes a long time to obtain assistance if a fall occurs. If the person doesn't get medical help within the first hour of an accident, they are more likely to die or develop chronic illnesses. Also, 50% of the elderly people who lay on the floor after a fall for a long time without medical assistance died within six months from the date of the accident. To overcome this and to assist the elderly a fall detection system that is more efficient has to be developed. This system should be inexpensive, portable, and accurate providing timely alerts to minimize the effects of delayed medical care reducing their fear of falling. Figure 1 shows the Fall Detection among Elderly People (FDEP) system that is generally followed to alert the caretakers and family members if any fall occurs. The available fall detection system is classified into various categories depending on the devices and intelligent algorithm involved: wearable sensor-based methods, non-wearable sensor-based methods and Artificial Intelligence (AI) based image or video processing methods such as machine learning and deep learning models. Non-wearable-based sensors are invasive and don’t solve the issue of adults who live alone. Non-wearable vision-based devices like cameras, suffers issue with privacy and high price even though they achieve efficiency and reliability. This kind of system needs to be installed physically at a place where the person lives, resulting in portability issue. Furthermore, half of all falls among elderly people occur outside the house. Hence it will be highly appreciable to have an AI based image or video processing fall detection system. Convolutional neural network has achieved great success in fields like object recognition and image classification with large training dataset. Hence deep learning could be applied for classifying elderly movements with sufficient data extracted from accelerometer and gyroscope end devices. This paper generated a signal-based image dataset, SimgFall from the existing accelerometer or gyroscope-based sensor data of the SiSFall dataset for early detection of fall to fasten the medical assistance process. Deep learning-based solution for fall detection is proposed for the generated SimgFall signal images and is trained using custom CNN and transfer learning models and the performance is analyzed. The proposed FallCNN model addresses the research problems as given below: Most of the existing research work explores the hardware-based wearable or non-wearable sensor data rather than AI-based architecture for fall detection. Upgradation for performance improvement has to be derived from the existing IoT models. Limitation in the affordability of the hardware models for fall detection has to be addressed Due to lack of fall dataset, it is very difficult to develop a deep learning solution for elderly fall detection problem. [ 23 ] The main objective of this research work is to propose a replacement strategy in which deep learning models can be trained using graph images generated from the SiSFall dataset's sensor data named SimgFall. Separate SimgFall images are created for each activity, which are then used to train and test the model (FallCNN and Transfer learning models). The proposed system examines four areas of behaviour: falling, jumping, walking, and stumbling. The major contribution of the proposed FallCNN system is Replacement strategy for hardware model adopted by most population is proposed and successfully devised using Deep learning model. This proposed unique custom CNN models like FallCNN_1, FallCNN_2, FallCNN_3 and FallCNN_4 deep learning architectures have been successfully implemented for the generated SimgFall dataset. The proposed model is compared with the two transfer learning models MobileNetV2 0.05 and MobileNetV2 0.1 all of which are trained and tested with the SimgFall dataset. To assist better learning for extracting elite patterns to be categorised, the FallCNN model’s included multiple convolution models with effective activation layers. The results for both custom FallCNN and pre-trained models with SimgFall dataset are shown in the form of ROC curve, accuracy and loss for both training and validation respectively. The remainder of the research paper is organized as follows: Section 2 summarises previous work in the field of fall detection, whereas Section 3 details the proposed work and architectures. The experimental analysis and results of custom FallCNN and transfer learning models are detailed in Section 4 . Section 5 concludes the work with the future scope. 2. Literature Review Pannurat et al [ 6 ] analysed fall detection based on wearable sensor-based methods and ambient sensor-based methods. Sensors such as gyroscope, pressure sensor, accelerometer, and microphone are used to detect the early fall detection in which the results of accelerometer are prominently and widely used among these. A technique based on near-field floor imaging sensors was proposed by Rimminen et al. [ 7 ] in 2010 where a two-state Markov chain model was used to classify the activities, and pose estimation was performed using Bayesian filtering. In 2011, Gjoreski [ 8 ] proposed a method that uses an accelerometer sensor placed on the chest, thigh, ankle, and waist to detect falls. The results showed that sensors placed on the chest or waist were effective in detecting falls, whereas the combined results also achieved best. Li et al. [ 9 ] detected falls by collecting the information from sensors mounted on furniture (bed and chair) along with the data obtained from the accelerometer placed on the person's thigh, wrist, ankle, chest, and forearm. The most commonly used ambient sensors are piezoelectric, acoustic sensors, infrared sensors, cameras, Kinect RGBD cameras, and so on [ 10 ]. Tzeng et al. [ 11 ] used an infrared camera to distinguish the subject's behaviour and a pressure sensor to detect the pressure on the floor. The camera has been widely used among ambient sensors, and the rapid growth of computer vision has boosted the use of video-based methods for fall detection [ 4 ]. Challenges faced during detection of fall in the small available population (elderly people), regardless of acquisition strategy, decreases their accuracy in real-world applications. Very few publicly accessible dataset that includes falls and activities of daily living (ADL) data is available for research [ 23 ]. Several research works [ 12 , 13 , 14 ] analysed and detected the motion of a person for prediction of fall. Since the self-developed embedded system can be easily replicated, SisFall data [ 14 ] is considered as the most accurate data for fall and ADL analysis among the available data. Computer assisted algorithm to detect fall has been widely researched using the rule-based methods, machine learning and Artificial Intelligence based models [ 5 ]. Most of the models analysed data obtained from wearable sensors using a thresholding technique or a set of rule-based predictions [ 15 ]. Sannino et al. [ 16 ] devised a rule based supervised method to analyse the accelerometer data which decided whether the given event is a fall or not. Logistic regression (LR), support vector machine (SVM), Neural networks (NN), multilayer perceptron’s (MLP), regularized discriminant classifier [ 19 , 20 ], nearest neighbours (KNN), naive Bayes classifier (NB) and random forest (RF) [ 17 ] are used for estimating fall risk where these approaches used the notable individual characteristics to access the fall. Also, inertial sensors are used to determine fall risk that uses sophisticated machine learning approaches, resulting in low classification accuracy [ 19 ]. If the system classifies a patient who has a high risk of falling as low risk, there is a possibility of severe consequences, deterring much-needed treatments. As a result, achieving high accuracy is difficult, necessitating the use of powerful deep learning techniques. Marcos et al. [ 22 ] presented a 2D CNN model using optical image attempting to integrate motion information via optical flow for efficient fall detection. Deep learning techniques such as convolutional neural networks (CNN) [ 21 ] have gained huge success within the last few years in fields such as natural language processing (NLP), object detection. Deep learning on the other hand, could learn an efficient representation (function) automatically, which prompted to use CNN to perform an efficient fall detection technique in this research work. 3. Proposed Work In this research work, custom CNN models in conjunction with transfer learning models have been analysed to detect the elderly movement. This research work proposed a unique approach within which models are trained by using signal images that are plotted from the sensor data of the SiSFall dataset named SimgFall. For each elderly movement like falling, jumping, walking, and stumble separate SimgFall images are generated, which are then used for training and testing the custom model. The proposed system architecture for Fall detection among elderly people are detailed in Fig. 2 . 3.1. Dataset SisFall dataset [ 14 ] includes sensor data detected using sensors like ADXL345, ITG3200, and MMA8451Q where ADXL345 and MMA8451Q are accelerometer sensors while ITG3200 is a gyro sensor. The dataset was generated with the assistance of 38 volunteers, divided into 2 groups, a group of elderly people and a group of young people. Young adults performed both falls and other activities, while some elderly people haven’t performed some activities due to some health issues. Multiple trials of the same activity like elderly movement like falling, jumping, walking, and stumbling were performed by the selected individuals are available in the SisFall dataset. 3.2. Generation of SimgFall Dataset Data pre-processing is an important process that has to be done in most data science-related research works. Pre-processing is commonly used to eliminate the null values, remove the unwanted rows or columns from the dataset. From the SisFall dataset, we have only considered 1992 CSV files because some volunteers haven’t performed all activities required, the CSV files of individuals who performed all the necessary activity were only considered for the experiment. From the selected CSV files, the first three columns of CSV file that correspond to accelerometer sensor data are selected and all the remaining columns were eliminated. The first three columns of each CSV file are plotted and saved as separate signal images into their corresponding class: fall, walk, jump and stumble. A total of 1992 signal images were generated from 1992 CSV files of the SisFall dataset named SimgFall . The SimgFall dataset is then divided into training and validation data and is fed to the CNN models. In the proposed work the sensor based SisFall data is converted to images to explore the fall detection in new perspective to improve the performance. The generated images are categorized into four categories: fall, walk, jump and stumble where each image scaled 256 ×256. 1992 images were captured from the SisFall dataset that includes 498 images per category. The dataset creation procedure is given in Algorithm 1 and sample of generated SimgFall images denoting each class is shown in Fig. 3 . Algorithm 1: SimgFall Dataset Creation 3.3. Convolutional Neural Networks (CNN) A convolutional neural network (CNN) consists of many neurons which work similarly to how neurons on the physical bodywork. A convolutional neural network includes an input layer, output layer, and several hidden layers. Generally, neurons in the CNN model are fed with input, followed by weighted sum computation, and forwarded to an activation function resulting in the output of the model. Each image is considered as an array initially, in the proposed scheme color image is considered as a 3D array that consists of red, blue, and green colors which have their level of intensities. A convolutional function is applied to the image data to generate a feature map using a failure detector and the convolutional operation is shown below. $${F}_{cov }\left(x,y\right)=\left(D*F\right)\left(x,y\right)= \sum _{i}\sum _{j}D\left(x+i,y+j\right)F(i,j)$$ \(wher\) e D is the input matrix representing the input image, F is the filter of size x and y and \({ F}_{cov }\) represents the output. In real life, humans can identify an object based on several features but might not look for all features for identifying an object. Similar is the case for CNN the object detector will only look at the necessary features. After the feature map is obtained a rectifier function is applied on top of it. Rectifier function ReLU (Rectified Linear Unit) is used since it increases the non-linearity. The rectifier act as a filter or function that breaks up the linearity. ReLU generates an output zero for the value less than zero or raw output otherwise. The mathematical illustration of the ReLU function is given in the equation below. $${R}_{cov }\left(x\right)=max(0, x)$$ After this step a pooling technique such as max-pooling/sum-pooling/mean-pooling would be applied on top of it to generate a pooled feature map so that the model would be easily able to identify towards which position the object is located in the image, the size and number of parameters is also reduced in this step and thus prevents the model from being overfitted. Generally, the next step is flattening, here the pooled features are converted into columns sequentially one after another so that they can be given to a neural network for further processing. More layers can be added and can be connected to a dense layer that includes ‘n’ output nodes depending on the number of classes. Each of the nodes in the final dense layer will be denoting the classes in the input given dataset. An example of the above-mentioned operations is shown in below Fig. 4 . More layers can be added and can be connected to a dense layer that includes ‘n’ output nodes depending on the number of classes. Each of the nodes in the final dense layer will be denoting the classes in the input given dataset. 3.3.1. Custom Model FallCNN_1 Model FallCNN_1 makes use of two 2-D convolutional layers and max-pooling layer, in which ReLU is employed as an activation function. The primary CNN layer uses 32, 3x3 essential filter processing the input data resulting in a 32-feature map. The max-pooling layer uses a kernel size of 2x2, which results in dimension reduction. Likewise to extract fine-scaled feature of the image, a second convolution layer with 32, 3x3 pixel filter, followed by a 2x2 kernel based max-pooling layer generating 14x14x32 features. A flattening layer is employed to convert this output into a vector which is further fed into a feed-forward neural network with 2 dense layers, with 128 hidden neurons and 4 output neurons. The activation function followed are ReLU for hidden layer and sigmoid for output layer. Categorical cross-entropy is employed as a loss function since the proposed analysis includes four categories. FallCNN_1 model achieved a classification accuracy of 94% with a loss of 0.264 by training 813,604 parameters. Figure 5 (a) shows the proposed flow diagram and (b) shows the summary of the FallCNN_1 model. FallCNN_2 Model FallCNN_1 is modified with respect to the pooling layer and a dropout layer is introduced resulting in FallCNN_2 model. FallCNN_2 uses the average-pooling technique whereas the FallCNN_1 model used the max-pooling technique. A dropout layer is used in this model to stop overfitting. A 20% drop allowed to prune the less conducive neurons that conjointly helps in reducing the generalization error. FallCNN_2 provided higher performance measures on the generated graph image dataset when compared with the FallCNN_1 model. The FallCNN_2 model achieved an accuracy of 95% with a loss of 0.118. Although the total parameters generated for this model and previous are same FallCNN_2 enhanced the accuracy by 1% due to average pooling and drop out function. Figures 6 (a) and (b) show the flow diagram and its model summary of the FallCNN_2 model. FallCNN_3 Model FallCNN_3 includes same number of convolution, pooling and dropout layers as of the previous one where the feature is remarkably increased from 62x62 to 220x220 in the convolution layers. The performance of FallCNN_3 model achieved an improved accuracy of 97% with a loss of 0.098 by analysing 11,954,724 features. The results shows that Fig. 7 (a) and (b) shows the flow diagram and model outline of the FallCNN_3 model. FallCNN_4 Model The FallCNN_4 model differs from the above-mentioned models in terms of layers and the input parameters used. FallCNN_4 consists of three main convolution blocks. Here each convolution block is followed by another intermediate block. After each intermediate convolution block, a max-pooling/average pooling and dropout layer with a 20% drop rate is used to prevent the model from overfitting. The same structure repeated for three times resulting in highest accuracy of 98% with a loss of 0.0833. FallCNN_4 model uses categorical cross-entropy as a loss function similar to the other models. Figure 8 (a) and (b) shows the flow diagram of FallCNN_4 model and its model summary. 3.3.2. Transfer Learning Models Transfer Learning model is used for applying the knowledge that is learned from a task to solve another almost similar task. In the proposed work we have created the models with the help of the Edge Impulse platform which is commonly used for machine learning on edge devices. Two transfer learning models: MobileNetV2 0.1 and MobileNetV2 0.05 are trained using the graphic images that are created with the help of the SiSFall dataset. MobileNetV2 is a lionvolutional Neural Netghtweight model which is also a convolutional neural network model which has the capability of classifying images into 1000 different categories. The two models MobileNetV2 0.1 and MobileNetV2 0. 05 considered for this research are designed in such a way that it has 10 neurons in its final layer with 0.1 dropouts. MobileNetV2 0.1 and MobileNetV2 0.05 achieved an accuracy of 97.2%, and 97.5% respectively. MobileNetV2 0. 05 has a slighter improved performance over the MobileNetV2 0.1 model. These transfer learning models had some problems in distinguishing some activities properly. 4. Results And Discussion This section presents a comprehensive view of the dataset, experimentation, model training, and validation. Performance comparison of the proposed approach with the existing works has also been presented in this section 4.1. Model Training and Validation The proposed FallCNN models and the transfer learning models are trained with the help of graph images that are generated from the SisFall dataset. The distribution of the samples into the training and validation dataset is in the ratio of 1591: 401, i.e., 1591 images for training and 401 images are used for testing purposes. Target data with 1591 images include 1,378 samples of Fall, 88 samples of the jump,89 samples of stumble, and 36 samples of walk class. Validation data with 401 images include 345 samples of fall, 23 samples of the jump, 18 samples of stumble, and 10 samples of walk class. The models were trained for 50 epochs. The model is evaluated on various classification metrics that include accuracy, confusion matrix, F1-score, recall, and precision. Table 1 shows the training and validation accuracy and loss of custom FallCNN models. The dynamics of the model can be evaluated for under-fit, over-fit, and good fit using the learning curves detailed in Table 1 . If the training loss curve in the learning remains flat regardless of how validation loss tends to decrease until the end of the training, the model is said to be under-fitted. If the plot of training loss decreases with experience and the validation loss curve decreases until a certain point and then starts increasing, the model is said to be over-fitted. Finally, a model is said to be a good fit when the training and validation losses stabilize and the difference between the two curves is negligible. It is observed from the above-resulted graphs that show good-fit behaviour within 50 epochs. 4.2. Effectiveness of the Proposed Architecture The effectiveness of the proposed model is analysed using various performance metrics like accuracy, loss, confusion matrix, recall, precision, and f1-score. With the help of these performance metrics, it can be clearly understood which model achieves better performance. The performance metrics of every model are detailed in this section. Accuracy Multiclass accuracy is the frequency with which the predicted value matches the actual value divided by the total prediction made. The expression for accuracy is given in Eq. ( 1 ). $$Accuracy = \frac{Count of Correctly predicted data points}{Total EquationNumber of data points}$$ 1 Precision and Recall The proportion of correctly predicted positive data is recalled whereas, precision is the proportion of true positives from all the data that are classified as positive by the network as given in (2) and (3). $$Precision = \frac{True Positive}{True positive + False positive}$$ 2 $$Recall = \frac{True positive }{True positive+ False negatve}$$ 3 F1-Score F1- score is the value obtained from a combination of precision and recall. From Eqs. ( 2 ) and ( 3 ) F1 score can be computed as given in (4). $$F =2*\frac{Precision*recall}{Precision + recall }$$ 4 4.2.1. Performance Measures for Four class classification of FallCNN Model This section shows the results of performance metrics such as accuracy, confusion matrix, loss, recall, precision, and f1-score on our custom CNN models. Table 2 shows the confusion matrix resulted for Fall, Jump, Stumble and Walk for FallCNN model. Table 3 details the Recall, Precision, F1 score, Loss and Accuracy obtained from the confusion matrix. Table 2. Confusion matrix for FallCNN models FallCNN_1 FallCNN_2 Fall Jump Stumble Walk Fall Jump Stumble Walk Fall 99% 0% 1% 0% 96% 3% 1% 0% Jump 48% 52% 0% 0% 9% 91% 0% 0% Stumble 22% 0% 78% 0% 4% 0% 92% 4% Walk 0% 0% 30% 70% 0% 0% 0% 100% FallCNN_3 FallCNN_4 Fall 100% 0% 0% 0% 99% 1% 0% 0% Jump 13% 87% 0% 0% 9% 91% 0% 0% Stumble 4% 0% 92% 4% 4% 0% 92% 4% Walk 0% 0% 1% 99% 0% 0% 0% 100% Table 3. Performance Summary for FallCNN models FallCNN_1 FallCNN_2 Fall Jump Stumble Walk Fall Jump Stumble Walk Recall 0.58 1 0.71 1 0.88 0.96 0.98 0.96 Precision 0.99 0.52 0.78 0.7 0.96 0.91 0.92 1 F1 Score 0.73 0.68 0.74 0.82 0.91 0.93 0.94 0.97 Loss 0.264 0.118 Accuracy 94% 95% FallCNN_3 FallCNN_4 Recall 0.85 1 0.98 0.96 0.88 0.98 1 0.96 Precision 1 0.87 0.92 0.99 0.99 0.91 0.92 1 F1 Score 0.91 0.93 0.94 0.97 0.93 0.94 0.95 0.97 Loss 0.098 0.083 Accuracy 97% 98% 4.2.2. Transfer Learning Models The result of performance metrics such as accuracy, confusion matrix, loss, recall, precision, and f1-score on the transfer learning models are shown in Table 4 and Table 5 . Table 4. Confusion matrix for MobileNetV2 models MobileNetV2 0.05 MobileNetV2 0.1 Fall Jump Stumble Walk Fall Jump Stumble Walk Fall 100% 0% 0% 0% 99% 1% 0% 0% Jump 34% 66% 0% 0% 14% 86% 0% 0% Stumble 6% 0% 94% 0% 0% 0% 92% 8% Walk 0% 0% 0% 100% 0% 0% 17% 83% Table 5. Performance summary for MobileNetV2 models MobileNetV2 0.05 MobileNetV2 0.1 Fall Jump Stumble Walk Fall Jump Stumble Walk Recall 0.71 1 1 1 0.87 0.98 0.84 0.91 Precision 1 0.66 0.94 1 0.99 0.86 0.92 0.83 F1 Score 0.83 0.79 0.96 1 0.92 0.91 0.87 0.86 Loss 0.19 0.07 Accuracy 97.5 97.2% 4.2.3 Overall performance Table 6 shows the overall validation accuracy (Accuracy, F1 Score, Recall, Precision) for the custom FallCNN models and the transfer learning models. 4.3. Feature Explorer The feature explorers for the transfer learning models MobileNetV2 0.05 and MobileNetV2 0.1 on the training data generated with the help of edge impulse platform are shown in Figs. 9 and 10 . 4.4. Visual Interpretation of The Trained Model Features Figure 11 shows the visualization of features extracted by the inner layers of the proposed FallCNN_4 during the training process. This visualization helps a lot in understanding how a model is interpreting the image internally. The salient features on the convolution and pooling layer of FallCNN_4 is visualized since this model achieved better results compared to other versions. 5. Conclusion Elderly Human Activity Recognition using computer vision has attained huge research attention. Recognizing the elderly movement like fall, Jump Stumble and Walk with the help of AI based intelligent analysis can be benefited for elderly monitoring. FDEP analysed various elderly fall detection on IoT-based accelerometer/gyroscope devices. To make FEDP more effective, a deep learning based custom FallCNN models are designed for enhanced eldely monitoring. A SimgFall graph images dataset are created by plotting the sensor data obtained from the SiSFall dataset for intelligent prediction of the selected activity such as fall, jump, stumble, and walk. The new dataset was tested with six models (four custom FallCNN models and two transfer learning models like MobileNetV2 0.05, MobileNetV2 0.1). In the case of transfer learning models, MobileNetV2 0.05 performed well when compared to the MobileNetV2 0.1. Similarly, FallCNN_4 provided better results among all the custom FallCNN models and the transfer learning models adopted in the proposed work. The performance metrics of the FallCNN_4 model with an accuracy of 98% outperforming all other Custom CNN (FallCNN) and Transfer learning models (MobileNetV2). Our future work includes (a) Working on real time elderly data collection more aligned with real-life scenarios, with more activities to analyse and predict. (b) Abnormal elderly behaviour prediction to evaluate their activity and to support health care services. Declarations Acknowledgements Not applicable. Author contributions M. Vergin Raja Sarobin and L. Jani Anbarasi formulated the work and M. Vergin Raja Sarobin and Jishnuraj k completed the implementation and M. Vergin Raja Sarobin , L. Jani Anbarasi and Jishnuraj. K wrote the main manuscript text , Rukmani P, S Graceline Jasmine and Modigari Narendra prepared the figures and validated the results. All authors reviewed the manuscript. Funding Not applicable. Availability of data and materials All data generated or analysed during this study are included in this published article. Ethics approval and consent to participate Not applicable. Consent for publication Not applicable. Competing interests The authors declare that they have no competing interests. References World Health Organization. WHO Global Report on Falls Prevention in Older Age; World Health Organization: Geneva, Switzerland, 2007. S. Lord, S. Smith, and J. Menant, “Vision and falls in older people: Risk factors and intervention strategies,” Clin. Geriatric Med., vol. 26, pp. 569–581 , 2010. Igual, R.; Medrano, C.; Plaza, I. Challenges, issues and trends in fall detection systems. BioMedical Engineering OnLine 2013, 12, 1–24. M.Mubashir, L. Shao, and L. Seed, “A survey on fall detection: Principles and approaches,” Neurocomputing, vol. 100, pp. 144–152 , 2013. L. Montesinos, R. Castaldo, and L. Pecchia, “Wearable inertial sensors for fall risk assessment and prediction in older adults: A systematic review and meta-analysis,” IEEE Trans. Neural Syst. Rehabil. Eng., vol. 26, no. 3, pp. 573–582, Mar. 2018. N. Pannurat, S. Thiemjarus, and E. Nantajeewarawat, “Automatic fall monitoring: A review,” Sensors, vol. 14, pp. 12900–12936 , 2014. Rimminen H., Lindström J., Linnavuo M., Sepponen R. Detection of falls among the elderly by a floor sensor using the electric near field . IEEE Trans. Inf. Technol. Biomed. 2010; 14 : 1475–1476 Gjoreski H., Lustrek M., Gams M. Accelerometer placement for posture recognition and fall detection. Proceedings of the Seventh International Conference on Intelligent Environments; Nottingham, UK. 25–28 July 2011; pp. 47–54. [Google Scholar] Li Q., Stankovic J.A. Grammar-based, posture- and context-cognitive detection for falls with different activity levels. Proceedings of the Second Conference on Wireless Health; La Jolla, CA, USA. 10–13 October 2011; pp. 1–10. E. E. Stone and M. Skubic, “Fall detection in homes of older adults using the Microsoft Kinect,” IEEE J. Biomed. Health Informat., vol. 19, no. 1, pp. 290–301, Jan. 2015. Tzeng H.-W., Chen M.-Y., Chen J.-Y. Design of fall detection system with floor pressure and infrared image. Proceedings of the International Conference on System Science and Engineering; Taipei, Taiwan. 1–3 July 2010; pp. 131–135. [Google Scholar] Vavoulas, G.; Pediaditis, M.; Chatzaki, C.; Spanakis, E.; Tsiknakis, M. The MobiFall Dataset: Fall Detection and Classification with a Smartphone. International Journal of Monitoring and Surveillance Technologies Research 2014, 2, 44–56 Medrano, C.; Igual, R.; Plaza, I.; Castro, M. Detecting Falls as Novelties in Acceleration Patterns Acquired with Smartphones . Plos One 2014, 9 , e94811 . Sucerquia A, López JD, Vargas-Bonilla JF. SisFall: A Fall and Movement Dataset. Sensors. 2017; 17(1). L. J. Kau and C. S. Chen, “A smart phone-based pocket fall accident detection, positioning, and rescue system,” IEEE J. Biomed. Health Informat., vol. 19, no. 1, pp. 44–56, Jan. 2015. G. Sannino, I. De Falco, and G. De Pietro, “A supervised approach to automatically extract a set of rules to support fall detection in an mHealth system,” Appl. Soft Comput., vol. 34, pp. 205–216 , 2015. Palmerini, L., Klenk, J., Becker, C. and Chiari, L., 2020. Accelerometer-based fall detection using machine learning: training and testing on real-world falls. Sensors, 20(22), p.6479. J. Silva et al., “Comparing machine learning approaches for fall risk assessment,” in Proc. 10th Int. Joint Conf. Biomed. Eng. Syst. Technol. , 2017, pp. 223–230. J. Howcroft, E. D. Lemaire, and J. Kofman, “Prospective elderly fall prediction by older-adult fall-risk modeling with feature selection,” Biomed. Signal Process. Control, vol. 43, pp. 320–328 , 2018. Rodrigues, T.B., Salgado, D.P., Cordeiro, M.C., Osterwald, K.M., Teodiano Filho, F.B., de Lucena Jr , V.F., Naves, E.L. and Murray, N., 2018. Fall detection system by machine learning framework for public health. Procedia Computer Science, 141, pp.358–365. Y. LeCun, Y. Bengio, and G. Hinton, “Deep learning,” Nature, vol. 521, pp. 436–444 , 2015. A. Nunez-Marcos, C.Azkune, and I. Arganda-Carreras, “Vision-based fall detection with convolutional neural networks,” Wireless Commun. Mobile Comput., vol. 2017, 2017, Art. no. 9474806. Lu, N., Wu, Y., Feng, L. and Song, J., 2018. Deep learning for fall detection: Three-dimensional CNN combined with LSTM on video kinematic data. IEEE journal of biomedical and health informatics, 23(1), pp.314–323. Tables Table 1 and 6 are available in the Supplementary Files section. Additional Declarations No competing interests reported. Supplementary Files Tables16.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-1884093","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":123980841,"identity":"eef33514-dcec-4be2-934d-40126f3d6785","order_by":0,"name":"M. Vergin Raja Sarobin","email":"","orcid":"","institution":"Vellore Institute of Technology","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"M.","middleName":"Vergin Raja","lastName":"Sarobin","suffix":""},{"id":123980843,"identity":"b155953d-b3c3-4a7c-9c24-f872de8247e0","order_by":1,"name":"Jishnuraj k","email":"","orcid":"","institution":"Vellore Institute of Technology","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Jishnuraj","middleName":"","lastName":"k","suffix":""},{"id":123980845,"identity":"28033fa8-bace-4966-9460-9420a9a17624","order_by":2,"name":"L. Jani Anbarasi","email":"","orcid":"","institution":"Vellore Institute of Technology","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"L.","middleName":"Jani","lastName":"Anbarasi","suffix":""},{"id":123980847,"identity":"9e4fbdde-b778-4e22-a2af-4cbb0b3a90ad","order_by":3,"name":"P Rukmani","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAz0lEQVRIiWNgGAWjYFACNhAhwcDA3sBwACzATLQWngMMBw6QoAWkK4EBag0BYD4jLXXDzx0WDOaSbwwPf2Cwk2dg58WvU+ZG2rGbvWckGCxn5xgAHZZs2MDMl4BXi4REetsN3jYJBoPbaQlALcwJDMw8BgS13PwL0nLzGEhLPTFa0o7dBttyg/kAUMthIrTwPEu7LdsmwWPZk3zgwBmD44ZtBLWwp5ndfNtWJ2fOfrD5Q0VFtTw//xn8WmAAarIBIp4IA+JMHgWjYBSMghEJAEJBPsXVFxSBAAAAAElFTkSuQmCC","orcid":"","institution":"Vellore Institute of Technology","correspondingAuthor":true,"submittingAuthor":false,"prefix":"","firstName":"P","middleName":"","lastName":"Rukmani","suffix":""},{"id":123980859,"identity":"dd2b39a1-86ea-4746-ba15-348141711e0a","order_by":4,"name":"S Graceline Jasmine","email":"","orcid":"","institution":"Vellore Institute of Technology","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"S","middleName":"Graceline","lastName":"Jasmine","suffix":""},{"id":123980860,"identity":"9d04d090-2aba-4ae3-a566-0e8ef1450ba5","order_by":5,"name":"Modigari Narendra","email":"","orcid":"","institution":"Vellore Institute of Technology","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Modigari","middleName":"","lastName":"Narendra","suffix":""}],"badges":[],"createdAt":"2022-07-22 05:59:06","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-1884093/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-1884093/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":24613020,"identity":"5dca12f2-9811-4e7f-8bd0-543147cdc3a6","added_by":"auto","created_at":"2022-08-01 17:36:50","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":32298,"visible":true,"origin":"","legend":"\u003cp\u003e\tFall Detection among Elderly People (FDEP) model\u0026nbsp;\u003c/p\u003e\u003cp\u003e\u003cbr\u003e\u003c/p\u003e","description":"","filename":"1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-1884093/v1/a7086bea30cc86af769f4a78.jpg"},{"id":24613378,"identity":"fb39a00b-e922-4bb3-aede-3b3524581dd5","added_by":"auto","created_at":"2022-08-01 17:41:50","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":144999,"visible":true,"origin":"","legend":"\u003cp\u003e\tThe architecture of Fall Detection of Elderly People (FDEP)\u003c/p\u003e\u003cp\u003e\u003cbr\u003e\u003c/p\u003e","description":"","filename":"2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-1884093/v1/ffd60b91d131dc99714c5a23.jpg"},{"id":24614407,"identity":"3f018b5e-57e2-4e85-9047-2958c0fbb84d","added_by":"auto","created_at":"2022-08-01 17:46:50","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":45283,"visible":true,"origin":"","legend":"\u003cp\u003eSample image for each activity of\u0026nbsp;\u003cstrong\u003eSimgFall \u003c/strong\u003edataset\u003c/p\u003e","description":"","filename":"3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-1884093/v1/8ef471de0328e18bc5fc2f0e.jpg"},{"id":24613382,"identity":"6d8a495d-b328-4d6d-9235-a92aed8a9e16","added_by":"auto","created_at":"2022-08-01 17:41:50","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":34293,"visible":true,"origin":"","legend":"\u003cp\u003eCNN model working example\u003c/p\u003e","description":"","filename":"4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-1884093/v1/2c8fbaf14a1a256bc6cbbb5f.jpg"},{"id":24613383,"identity":"47330086-0826-4ad9-8c73-cfb88338712c","added_by":"auto","created_at":"2022-08-01 17:41:50","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":68342,"visible":true,"origin":"","legend":"\u003cp\u003e(a) FallCNN_1 Model (b) Summary of FallCNN_1 Model\u003c/p\u003e\u003cp\u003e\u003cbr\u003e\u003c/p\u003e","description":"","filename":"5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-1884093/v1/be98f978a9682473aea8aca1.jpg"},{"id":24613023,"identity":"0c82f6c4-fd03-45ac-bc43-7fee5605b5b8","added_by":"auto","created_at":"2022-08-01 17:36:50","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":64198,"visible":true,"origin":"","legend":"\u003cp\u003e\t(a) FallCNN_2 Model\u0026nbsp;(b) Summary of FallCNN_2 Model\u003c/p\u003e\u003cp\u003e\u003cbr\u003e\u003c/p\u003e","description":"","filename":"6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-1884093/v1/2e34007a36c9d481d961be39.jpg"},{"id":24613379,"identity":"a443345f-e1c8-4f4f-b73c-b4e4fd118539","added_by":"auto","created_at":"2022-08-01 17:41:50","extension":"jpg","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":63240,"visible":true,"origin":"","legend":"\u003cp\u003e\t(a) FallCNN_3 Model (b) Summary of FallCNN_3 Model\u003c/p\u003e\u003cp\u003e\u003cbr\u003e\u003c/p\u003e","description":"","filename":"7.jpg","url":"https://assets-eu.researchsquare.com/files/rs-1884093/v1/9a64e22f2dde4068853bad2f.jpg"},{"id":24613036,"identity":"efeceab3-cc10-4c95-911c-23702e87c6ae","added_by":"auto","created_at":"2022-08-01 17:36:50","extension":"jpg","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":229367,"visible":true,"origin":"","legend":"\u003cp\u003e(a) FallCNN_4 Model\u0026nbsp;(b) Summary of FallCNN_4 Model\u003c/p\u003e","description":"","filename":"8.jpg","url":"https://assets-eu.researchsquare.com/files/rs-1884093/v1/360bd75959baef9e473e0621.jpg"},{"id":24613381,"identity":"2e5e2bfe-9750-4a5f-897f-6a3f0c8c7b12","added_by":"auto","created_at":"2022-08-01 17:41:50","extension":"jpg","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":34970,"visible":true,"origin":"","legend":"\u003cp\u003eFeature explorer for MobileNetV2 0.1\u003c/p\u003e\u003cp\u003e\u003cbr\u003e\u003c/p\u003e","description":"","filename":"9.jpg","url":"https://assets-eu.researchsquare.com/files/rs-1884093/v1/3ffa7f07c32755f2dc817148.jpg"},{"id":24613062,"identity":"dfcda0b0-78a9-470d-ab04-7a99b64b1565","added_by":"auto","created_at":"2022-08-01 17:36:50","extension":"jpg","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":33323,"visible":true,"origin":"","legend":"\u003cp\u003eFeature explorer for MobileNetV2 0.05\u003c/p\u003e\u003cp\u003e\u003cbr\u003e\u003c/p\u003e","description":"","filename":"10.jpg","url":"https://assets-eu.researchsquare.com/files/rs-1884093/v1/50c082d942d2d2a577efbac2.jpg"},{"id":24613026,"identity":"845ea8e4-c19c-4b5a-8b28-a22c53393bd2","added_by":"auto","created_at":"2022-08-01 17:36:50","extension":"jpg","order_by":11,"title":"Figure 11","display":"","copyAsset":false,"role":"figure","size":98804,"visible":true,"origin":"","legend":"\u003cp\u003eVisualization of the most salient features on the convolution and pooling layers.\u003c/p\u003e","description":"","filename":"11.jpg","url":"https://assets-eu.researchsquare.com/files/rs-1884093/v1/51cba322165e1fef6223af41.jpg"},{"id":30292468,"identity":"e614466a-a8e8-423d-9cce-188358ee86c4","added_by":"auto","created_at":"2022-12-14 02:29:31","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1077505,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-1884093/v1/239d00a8-4f72-4890-a815-df33b13766fd.pdf"},{"id":24613022,"identity":"50fa92a2-7c2f-45df-a799-46add9a92695","added_by":"auto","created_at":"2022-08-01 17:36:50","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":2222826,"visible":true,"origin":"","legend":"","description":"","filename":"Tables16.docx","url":"https://assets-eu.researchsquare.com/files/rs-1884093/v1/0b2ccd48771f0c9457915adf.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Fall Detection among Elderly Person using FallCNN and Transfer Learning Models","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eIn today\u0026apos;s world, one of the most common causes of death is fall which is more common among elderly people due to their physical weakness. Research done by World Health Organization (WHO) suggests that 42% of people above the age of 65 suffer from at least one fall. So as age increases elderly fall rates also increase. Although, there are a lot of fall detection devices available in the market most of them are expensive, non-portable, or inaccurate.\u003c/p\u003e\n\u003cp\u003eThe fear of falling is more among people who live alone because it takes a long time to obtain assistance if a fall occurs. If the person doesn\u0026apos;t get medical help within the first hour of an accident, they are more likely to die or develop chronic illnesses. Also, 50% of the elderly people who lay on the floor after a fall for a long time without medical assistance died within six months from the date of the accident. To overcome this and to assist the elderly a fall detection system that is more efficient has to be developed. This system should be inexpensive, portable, and accurate providing timely alerts to minimize the effects of delayed medical care reducing their fear of falling.\u003c/p\u003e\n\u003cp\u003eFigure \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e shows the Fall Detection among Elderly People (FDEP) system that is generally followed to alert the caretakers and family members if any fall occurs. The available fall detection system is classified into various categories depending on the devices and intelligent algorithm involved: wearable sensor-based methods, non-wearable sensor-based methods and Artificial Intelligence (AI) based image or video processing methods such as machine learning and deep learning models. Non-wearable-based sensors are invasive and don\u0026rsquo;t solve the issue of adults who live alone. Non-wearable vision-based devices like cameras, suffers issue with privacy and high price even though they achieve efficiency and reliability. This kind of system needs to be installed physically at a place where the person lives, resulting in portability issue. Furthermore, half of all falls among elderly people occur outside the house. Hence it will be highly appreciable to have an AI based image or video processing fall detection system.\u003c/p\u003e\n\u003cp\u003eConvolutional neural network has achieved great success in fields like object recognition and image classification with large training dataset. Hence deep learning could be applied for classifying elderly movements with sufficient data extracted from accelerometer and gyroscope end devices. This paper generated a signal-based image dataset, SimgFall from the existing accelerometer or gyroscope-based sensor data of the SiSFall dataset for early detection of fall to fasten the medical assistance process. Deep learning-based solution for fall detection is proposed for the generated SimgFall signal images and is trained using custom CNN and transfer learning models and the performance is analyzed.\u003c/p\u003e\n\u003cp\u003eThe proposed FallCNN model addresses the research problems as given below:\u003c/p\u003e\n\u003col\u003e\n \u003cli\u003eMost of the existing research work explores the hardware-based wearable or non-wearable sensor data rather than AI-based architecture for fall detection.\u003c/li\u003e\n \u003cli\u003e\u003cspan\u003e\n \u003cp\u003eUpgradation for performance improvement has to be derived from the existing IoT models.\u003c/p\u003e\n \u003c/span\u003e\u003c/li\u003e\n \u003cli\u003eLimitation in the affordability of the hardware models for fall detection has to be addressed\u003cspan\u003e\u0026nbsp;\u003c/span\u003e\u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eDue to lack of fall dataset, it is very difficult to develop a deep learning solution for elderly fall detection problem. [\u003cspan class=\"CitationRef\"\u003e23\u003c/span\u003e]\u003c/p\u003e\n \u003c/li\u003e\n\u003c/ol\u003e\n\n\u003cp\u003eThe main objective of this research work is to propose a replacement strategy in which deep learning models can be trained using graph images generated from the SiSFall dataset\u0026apos;s sensor data named SimgFall. Separate SimgFall images are created for each activity, which are then used to train and test the model (FallCNN and Transfer learning models). The proposed system examines four areas of behaviour: falling, jumping, walking, and stumbling. The major contribution of the proposed FallCNN system is\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003e\n \u003cp\u003eReplacement strategy for hardware model adopted by most population is proposed and successfully devised using Deep learning model.\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eThis proposed unique custom CNN models like FallCNN_1, FallCNN_2, FallCNN_3 and FallCNN_4 deep learning architectures have been successfully implemented for the generated SimgFall dataset.\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eThe proposed model is compared with the two transfer learning models MobileNetV2 0.05 and MobileNetV2 0.1 all of which are trained and tested with the SimgFall dataset.\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eTo assist better learning for extracting elite patterns to be categorised, the FallCNN model\u0026rsquo;s included multiple convolution models with effective activation layers.\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eThe results for both custom FallCNN and pre-trained models with SimgFall dataset are shown in the form of ROC curve, accuracy and loss for both training and validation respectively.\u003c/p\u003e\u003cbr\u003e\n \u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eThe remainder of the research paper is organized as follows: Section 2 summarises previous work in the field of fall detection, whereas Section \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e details the proposed work and architectures. The experimental analysis and results of custom FallCNN and transfer learning models are detailed in Section \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e. Section \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003e concludes the work with the future scope.\u003c/p\u003e"},{"header":"2. Literature Review","content":"\u003cp\u003ePannurat et al [\u003cspan class=\"CitationRef\"\u003e6\u003c/span\u003e] analysed fall detection based on wearable sensor-based methods and ambient sensor-based methods. Sensors such as gyroscope, pressure sensor, accelerometer, and microphone are used to detect the early fall detection in which the results of accelerometer are prominently and widely used among these.\u003c/p\u003e\n\u003cp\u003eA technique based on near-field floor imaging sensors was proposed by Rimminen et al. [\u003cspan class=\"CitationRef\"\u003e7\u003c/span\u003e] in 2010 where a two-state Markov chain model was used to classify the activities, and pose estimation was performed using Bayesian filtering. In 2011, Gjoreski [\u003cspan class=\"CitationRef\"\u003e8\u003c/span\u003e] proposed a method that uses an accelerometer sensor placed on the chest, thigh, ankle, and waist to detect falls. The results showed that sensors placed on the chest or waist were effective in detecting falls, whereas the combined results also achieved best. Li et al. [\u003cspan class=\"CitationRef\"\u003e9\u003c/span\u003e] detected falls by collecting the information from sensors mounted on furniture (bed and chair) along with the data obtained from the accelerometer placed on the person\u0026apos;s thigh, wrist, ankle, chest, and forearm.\u003c/p\u003e\n\u003cp\u003eThe most commonly used ambient sensors are piezoelectric, acoustic sensors, infrared sensors, cameras, Kinect RGBD cameras, and so on [\u003cspan class=\"CitationRef\"\u003e10\u003c/span\u003e]. Tzeng et al. [\u003cspan class=\"CitationRef\"\u003e11\u003c/span\u003e] used an infrared camera to distinguish the subject\u0026apos;s behaviour and a pressure sensor to detect the pressure on the floor. The camera has been widely used among ambient sensors, and the rapid growth of computer vision has boosted the use of video-based methods for fall detection [\u003cspan class=\"CitationRef\"\u003e4\u003c/span\u003e]. Challenges faced during detection of fall in the small available population (elderly people), regardless of acquisition strategy, decreases their accuracy in real-world applications. Very few publicly accessible dataset that includes falls and activities of daily living (ADL) data is available for research [\u003cspan class=\"CitationRef\"\u003e23\u003c/span\u003e]. Several research works [\u003cspan class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan class=\"CitationRef\"\u003e14\u003c/span\u003e] analysed and detected the motion of a person for prediction of fall. Since the self-developed embedded system can be easily replicated, SisFall data [\u003cspan class=\"CitationRef\"\u003e14\u003c/span\u003e] is considered as the most accurate data for fall and ADL analysis among the available data.\u003c/p\u003e\n\u003cp\u003eComputer assisted algorithm to detect fall has been widely researched using the rule-based methods, machine learning and Artificial Intelligence based models [\u003cspan class=\"CitationRef\"\u003e5\u003c/span\u003e]. Most of the models analysed data obtained from wearable sensors using a thresholding technique or a set of rule-based predictions [\u003cspan class=\"CitationRef\"\u003e15\u003c/span\u003e]. Sannino et al. [\u003cspan class=\"CitationRef\"\u003e16\u003c/span\u003e] devised a rule based supervised method to analyse the accelerometer data which decided whether the given event is a fall or not. Logistic regression (LR), support vector machine (SVM), Neural networks (NN), multilayer perceptron\u0026rsquo;s (MLP), regularized discriminant classifier [\u003cspan class=\"CitationRef\"\u003e19\u003c/span\u003e, \u003cspan class=\"CitationRef\"\u003e20\u003c/span\u003e], nearest neighbours (KNN), naive Bayes classifier (NB) and random forest (RF) [\u003cspan class=\"CitationRef\"\u003e17\u003c/span\u003e] are used for estimating fall risk where these approaches used the notable individual characteristics to access the fall.\u003c/p\u003e\n\u003cp\u003eAlso, inertial sensors are used to determine fall risk that uses sophisticated machine learning approaches, resulting in low classification accuracy [\u003cspan class=\"CitationRef\"\u003e19\u003c/span\u003e]. If the system classifies a patient who has a high risk of falling as low risk, there is a possibility of severe consequences, deterring much-needed treatments. As a result, achieving high accuracy is difficult, necessitating the use of powerful deep learning techniques. Marcos et al. [\u003cspan class=\"CitationRef\"\u003e22\u003c/span\u003e] presented a 2D CNN model using optical image attempting to integrate motion information via optical flow for efficient fall detection. Deep learning techniques such as convolutional neural networks (CNN) [\u003cspan class=\"CitationRef\"\u003e21\u003c/span\u003e] have gained huge success within the last few years in fields such as natural language processing (NLP), object detection. Deep learning on the other hand, could learn an efficient representation (function) automatically, which prompted to use CNN to perform an efficient fall detection technique in this research work.\u003c/p\u003e"},{"header":"3. Proposed Work","content":"\u003cp\u003eIn this research work, custom CNN models in conjunction with transfer learning models have been analysed to detect the elderly movement. This research work proposed a unique approach within which models are trained by using signal images that are plotted from the sensor data of the SiSFall dataset named SimgFall. For each elderly movement like falling, jumping, walking, and stumble separate SimgFall images are generated, which are then used for training and testing the custom model. The proposed system architecture for Fall detection among elderly people are detailed in Fig. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e\n\u003cdiv class=\"Section2\" id=\"Sec3\"\u003e\n \u003ch2\u003e3.1. Dataset\u003c/h2\u003e\n \u003cp\u003eSisFall dataset [\u003cspan class=\"CitationRef\"\u003e14\u003c/span\u003e] includes sensor data detected using sensors like ADXL345, ITG3200, and MMA8451Q where ADXL345 and MMA8451Q are accelerometer sensors while ITG3200 is a gyro sensor. The dataset was generated with the assistance of 38 volunteers, divided into 2 groups, a group of elderly people and a group of young people. Young adults performed both falls and other activities, while some elderly people haven\u0026rsquo;t performed some activities due to some health issues. Multiple trials of the same activity like elderly movement like falling, jumping, walking, and stumbling were performed by the selected individuals are available in the SisFall dataset.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv class=\"Section2\" id=\"Sec4\"\u003e\n \u003ch2\u003e3.2. Generation of SimgFall Dataset\u003c/h2\u003e\n \u003cp\u003eData pre-processing is an important process that has to be done in most data science-related research works. Pre-processing is commonly used to eliminate the null values, remove the unwanted rows or columns from the dataset. From the SisFall dataset, we have only considered 1992 CSV files because some volunteers haven\u0026rsquo;t performed all activities required, the CSV files of individuals who performed all the necessary activity were only considered for the experiment. From the selected CSV files, the first three columns of CSV file that correspond to accelerometer sensor data are selected and all the remaining columns were eliminated. The first three columns of each CSV file are plotted and saved as separate signal images into their corresponding class: fall, walk, jump and stumble. A total of 1992 signal images were generated from 1992 CSV files of the SisFall dataset named \u003cstrong\u003eSimgFall\u003c/strong\u003e. The \u003cstrong\u003eSimgFall\u003c/strong\u003e dataset is then divided into training and validation data and is fed to the CNN models. In the proposed work the sensor based SisFall data is converted to images to explore the fall detection in new perspective to improve the performance. The generated images are categorized into four categories: fall, walk, jump and stumble where each image scaled 256 \u0026times;256. 1992 images were captured from the SisFall dataset that includes 498 images per category. The dataset creation procedure is given in Algorithm \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e and sample of generated \u003cstrong\u003eSimgFall\u003c/strong\u003e images denoting each class is shown in Fig. \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e.\u003c/p\u003e\n \u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eAlgorithm 1:\u0026nbsp;\u003c/strong\u003e \u003cstrong style='font-weight: 700; color: rgb(0, 0, 0); font-family: \"Times New Roman\"; font-size: medium; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; letter-spacing: normal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none; white-space: normal; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial;'\u003eSimgFall\u003c/strong\u003e \u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003eDataset Creation\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cimg src=\"data:image/png;base64,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\"\u003e\u003c/strong\u003e\u003cbr\u003e\u003c/p\u003e\n \u003ch2\u003e\u003cstrong\u003e3.3. Convolutional Neural Networks (CNN)\u003c/strong\u003e\u003c/h2\u003e\n\u003c/div\u003e\n\u003cdiv class=\"Section2\" id=\"Sec5\"\u003e\n \u003cp\u003eA convolutional neural network (CNN) consists of many neurons which work similarly to how neurons on the physical bodywork. A convolutional neural network includes an input layer, output layer, and several hidden layers. Generally, neurons in the CNN model are fed with input, followed by weighted sum computation, and forwarded to an activation function resulting in the output of the model. Each image is considered as an array initially, in the proposed scheme color image is considered as a 3D array that consists of red, blue, and green colors which have their level of intensities. A convolutional function is applied to the image data to generate a feature map using a failure detector and the convolutional operation is shown below.\u003c/p\u003e\n \u003cdiv class=\"Equation\" id=\"Equa\"\u003e\n \u003cdiv class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e$${F}_{cov }\\left(x,y\\right)=\\left(D*F\\right)\\left(x,y\\right)= \\sum _{i}\\sum _{j}D\\left(x+i,y+j\\right)F(i,j)$$\u003c/div\u003e\n \u003c/div\u003e\n \u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n \u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u0026nbsp;\u003cspan class=\"mathinline\"\u003e\\(wher\\)\u003c/span\u003e\u0026nbsp;\u003c/span\u003ee D is the input matrix representing the input image, F is the filter of size x and y and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\({ F}_{cov }\\)\u003c/span\u003e\u003c/span\u003erepresents the output. In real life, humans can identify an object based on several features but might not look for all features for identifying an object. Similar is the case for CNN the object detector will only look at the necessary features. After the feature map is obtained a rectifier function is applied on top of it. Rectifier function ReLU (Rectified Linear Unit) is used since it increases the non-linearity. The rectifier act as a filter or function that breaks up the linearity. ReLU generates an output zero for the value less than zero or raw output otherwise. The mathematical illustration of the ReLU function is given in the equation below.\u003c/p\u003e\n \u003cdiv class=\"Equation\" id=\"Equb\"\u003e\n \u003cdiv class=\"mathdisplay\" id=\"FileID_Equb\" name=\"EquationSource\"\u003e$${R}_{cov }\\left(x\\right)=max(0, x)$$\u003c/div\u003e\n \u003c/div\u003e\n \u003cp\u003eAfter this step a pooling technique such as max-pooling/sum-pooling/mean-pooling would be applied on top of it to generate a pooled feature map so that the model would be easily able to identify towards which position the object is located in the image, the size and number of parameters is also reduced in this step and thus prevents the model from being overfitted. Generally, the next step is flattening, here the pooled features are converted into columns sequentially one after another so that they can be given to a neural network for further processing.\u003c/p\u003e\n \u003cp\u003eMore layers can be added and can be connected to a dense layer that includes \u0026lsquo;n\u0026rsquo; output nodes depending on the number of classes. Each of the nodes in the final dense layer will be denoting the classes in the input given dataset. An example of the above-mentioned operations is shown in below Fig. \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e.\u003c/p\u003e\n \u003cp\u003eMore layers can be added and can be connected to a dense layer that includes \u0026lsquo;n\u0026rsquo; output nodes depending on the number of classes. Each of the nodes in the final dense layer will be denoting the classes in the input given dataset.\u003c/p\u003e\n \u003cdiv class=\"Section3\" id=\"Sec6\"\u003e\n \u003ch2\u003e3.3.1. Custom Model\u003c/h2\u003e\n \u003cp\u003e\u003cstrong\u003eFallCNN_1 Model\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003eFallCNN_1 makes use of two 2-D convolutional layers and max-pooling layer, in which ReLU is employed as an activation function. The primary CNN layer uses 32, 3x3 essential filter processing the input data resulting in a 32-feature map. The max-pooling layer uses a kernel size of 2x2, which results in dimension reduction. Likewise to extract fine-scaled feature of the image, a second convolution layer with 32, 3x3 pixel filter, followed by a 2x2 kernel based max-pooling layer generating 14x14x32 features. A flattening layer is employed to convert this output into a vector which is further fed into a feed-forward neural network with 2 dense layers, with 128 hidden neurons and 4 output neurons. The activation function followed are ReLU for hidden layer and sigmoid for output layer. Categorical cross-entropy is employed as a loss function since the proposed analysis includes four categories. FallCNN_1 model achieved a classification accuracy of 94% with a loss of 0.264 by training 813,604 parameters. Figure \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003e (a) shows the proposed flow diagram and (b) shows the summary of the FallCNN_1 model.\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eFallCNN_2 Model\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003eFallCNN_1 is modified with respect to the pooling layer and a dropout layer is introduced resulting in FallCNN_2 model. FallCNN_2 uses the average-pooling technique whereas the FallCNN_1 model used the max-pooling technique. A dropout layer is used in this model to stop overfitting. A 20% drop allowed to prune the less conducive neurons that conjointly helps in reducing the generalization error. FallCNN_2 provided higher performance measures on the generated graph image dataset when compared with the FallCNN_1 model. The FallCNN_2 model achieved an accuracy of 95% with a loss of 0.118. Although the total parameters generated for this model and previous are same FallCNN_2 enhanced the accuracy by 1% due to average pooling and drop out function. Figures \u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003e (a) and (b) show the flow diagram and its model summary of the FallCNN_2 model.\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eFallCNN_3 Model\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003eFallCNN_3 includes same number of convolution, pooling and dropout layers as of the previous one where the feature is remarkably increased from 62x62 to 220x220 in the convolution layers. The performance of FallCNN_3 model achieved an improved accuracy of 97% with a loss of 0.098 by analysing 11,954,724 features. The results shows that Fig. \u003cspan class=\"InternalRef\"\u003e7\u003c/span\u003e (a) and (b) shows the flow diagram and model outline of the FallCNN_3 model.\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eFallCNN_4 Model\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003eThe FallCNN_4 model differs from the above-mentioned models in terms of layers and the input parameters used. FallCNN_4 consists of three main convolution blocks. Here each convolution block is followed by another intermediate block. After each intermediate convolution block, a max-pooling/average pooling and dropout layer with a 20% drop rate is used to prevent the model from overfitting. The same structure repeated for three times resulting in highest accuracy of 98% with a loss of 0.0833. FallCNN_4 model uses categorical cross-entropy as a loss function similar to the other models. Figure 8 (a) and (b) shows the flow diagram of FallCNN_4 model and its model summary.\u003c/p\u003e\n \u003c/div\u003e\n \u003cdiv class=\"Section3\" id=\"Sec7\"\u003e\n \u003ch2\u003e3.3.2. Transfer Learning Models\u003c/h2\u003e\n \u003cp\u003eTransfer Learning model is used for applying the knowledge that is learned from a task to solve another almost similar task. In the proposed work we have created the models with the help of the Edge Impulse platform which is commonly used for machine learning on edge devices. Two transfer learning models: MobileNetV2 0.1 and MobileNetV2 0.05 are trained using the graphic images that are created with the help of the SiSFall dataset. MobileNetV2 is a lionvolutional Neural Netghtweight model which is also a convolutional neural network model which has the capability of classifying images into 1000 different categories. The two models MobileNetV2 0.1 and MobileNetV2 0. 05 considered for this research are designed in such a way that it has 10 neurons in its final layer with 0.1 dropouts. MobileNetV2 0.1 and MobileNetV2 0.05 achieved an accuracy of 97.2%, and 97.5% respectively. MobileNetV2 0. 05 has a slighter improved performance over the MobileNetV2 0.1 model. These transfer learning models had some problems in distinguishing some activities properly.\u003c/p\u003e\n \u003c/div\u003e\n\u003c/div\u003e"},{"header":"4. Results And Discussion","content":"\u003cp\u003eThis section presents a comprehensive view of the dataset, experimentation, model training, and validation. Performance comparison of the proposed approach with the existing works has also been presented in this section\u003c/p\u003e\n\u003cdiv class=\"Section2\" id=\"Sec9\"\u003e\n \u003ch2\u003e4.1. Model Training and Validation\u003c/h2\u003e\n \u003cp\u003eThe proposed FallCNN models and the transfer learning models are trained with the help of graph images that are generated from the SisFall dataset. The distribution of the samples into the training and validation dataset is in the ratio of 1591: 401, i.e., 1591 images for training and 401 images are used for testing purposes. Target data with 1591 images include 1,378 samples of Fall, 88 samples of the jump,89 samples of stumble, and 36 samples of walk class. Validation data with 401 images include 345 samples of fall, 23 samples of the jump, 18 samples of stumble, and 10 samples of walk class. The models were trained for 50 epochs. The model is evaluated on various classification metrics that include accuracy, confusion matrix, F1-score, recall, and precision. Table\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e shows the training and validation accuracy and loss of custom FallCNN models.\u003c/p\u003e\n \u003cp\u003eThe dynamics of the model can be evaluated for under-fit, over-fit, and good fit using the learning curves detailed in Table\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e. If the training loss curve in the learning remains flat regardless of how validation loss tends to decrease until the end of the training, the model is said to be under-fitted. If the plot of training loss decreases with experience and the validation loss curve decreases until a certain point and then starts increasing, the model is said to be over-fitted. Finally, a model is said to be a good fit when the training and validation losses stabilize and the difference between the two curves is negligible. It is observed from the above-resulted graphs that show good-fit behaviour within 50 epochs.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv class=\"Section2\" id=\"Sec10\"\u003e\n \u003ch2\u003e4.2. Effectiveness of the Proposed Architecture\u003c/h2\u003e\n \u003cp\u003eThe effectiveness of the proposed model is analysed using various performance metrics like accuracy, loss, confusion matrix, recall, precision, and f1-score. With the help of these performance metrics, it can be clearly understood which model achieves better performance. The performance metrics of every model are detailed in this section.\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eAccuracy\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003eMulticlass accuracy is the frequency with which the predicted value matches the actual value divided by the total prediction made. The expression for accuracy is given in Eq.\u0026nbsp;(\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e\n \u003cdiv class=\"Equation\" id=\"Equ1\"\u003e\n \u003cdiv class=\"mathdisplay\" id=\"FileID_Equ1\" name=\"EquationSource\"\u003e$$Accuracy = \\frac{Count of Correctly predicted data points}{Total EquationNumber of data points}$$\u003c/div\u003e\n \u003cdiv class=\"EquationNumber\"\u003e1\u003c/div\u003e\n \u003c/div\u003e\n \u003cp\u003e\u003cstrong\u003ePrecision and Recall\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003eThe proportion of correctly predicted positive data is recalled whereas, precision is the proportion of true positives from all the data that are classified as positive by the network as given in (2) and (3).\u003c/p\u003e\n \u003cdiv class=\"Equation\" id=\"Equ2\"\u003e\n \u003cdiv class=\"mathdisplay\" id=\"FileID_Equ2\" name=\"EquationSource\"\u003e$$Precision = \\frac{True Positive}{True positive + False positive}$$\u003c/div\u003e\n \u003cdiv class=\"EquationNumber\"\u003e2\u003c/div\u003e\n \u003c/div\u003e\n \u003cdiv class=\"Equation\" id=\"Equ3\"\u003e\n \u003cdiv class=\"mathdisplay\" id=\"FileID_Equ3\" name=\"EquationSource\"\u003e$$Recall = \\frac{True positive }{True positive+ False negatve}$$\u003c/div\u003e\n \u003cdiv class=\"EquationNumber\"\u003e3\u003c/div\u003e\n \u003c/div\u003e\n \u003cp\u003e\u003cstrong\u003eF1-Score\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003eF1- score is the value obtained from a combination of precision and recall. From Eqs.\u0026nbsp;(\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e) and (\u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e) F1 score can be computed as given in (4).\u003c/p\u003e\n \u003cdiv class=\"Equation\" id=\"Equ4\"\u003e\n \u003cdiv class=\"mathdisplay\" id=\"FileID_Equ4\" name=\"EquationSource\"\u003e$$F =2*\\frac{Precision*recall}{Precision + recall }$$\u003c/div\u003e\n \u003cdiv class=\"EquationNumber\"\u003e4\u003c/div\u003e\n \u003c/div\u003e\n \u003cdiv class=\"Section3\" id=\"Sec11\"\u003e\n \u003ch2\u003e4.2.1. Performance Measures for Four class classification of FallCNN Model\u003c/h2\u003e\n \u003cp\u003eThis section shows the results of performance metrics such as accuracy, confusion matrix, loss, recall, precision, and f1-score on our custom CNN models. Table \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e shows the confusion matrix resulted for Fall, Jump, Stumble and Walk for FallCNN model. Table \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e details the Recall, Precision, F1 score, Loss and Accuracy obtained from the confusion matrix.\u003c/p\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:8.0pt;margin-left:0in;line-height:107%;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:justify;'\u003e\u003cspan style='font-size:16px;line-height:107%;font-family:\"Times New Roman\",serif;'\u003eTable 2.\u0026nbsp;\u003c/span\u003e\u003cspan style='font-size:16px;line-height:107%;font-family:\"Times New Roman\",serif;'\u003e\u0026nbsp;Confusion matrix for FallCNN models\u003c/span\u003e\u003c/p\u003e\n \u003ctable style=\"border-collapse:collapse;border:none;\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 62.75pt;border: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" style=\"width: 194.2pt;border-top: 1pt solid windowtext;border-right: 1pt solid windowtext;border-bottom: 1pt solid windowtext;border-image: initial;border-left: none;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eFallCNN_1\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" style=\"width: 193.85pt;border-top: 1pt solid windowtext;border-right: 1pt solid windowtext;border-bottom: 1pt solid windowtext;border-image: initial;border-left: none;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eFallCNN_2\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eFall\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.7pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eJump\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eStumble\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.8pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eWalk\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.15pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eFall\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.8pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eJump\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eStumble\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 48.05pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eWalk\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 62.75pt;border-right: 1pt solid windowtext;border-bottom: 1pt solid windowtext;border-left: 1pt solid windowtext;border-image: initial;border-top: none;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eFall\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(0, 176, 80);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e99%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.7pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: red;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e1%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.8pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.15pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(0, 176, 80);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e96%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.8pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: red;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e3%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: red;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e1%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 48.05pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 62.75pt;border-right: 1pt solid windowtext;border-bottom: 1pt solid windowtext;border-left: 1pt solid windowtext;border-image: initial;border-top: none;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eJump\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: red;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e48%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.7pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(0, 176, 80);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e52%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.8pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.15pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: red;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e9%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.8pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(0, 176, 80);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e91%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 48.05pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 62.75pt;border-right: 1pt solid windowtext;border-bottom: 1pt solid windowtext;border-left: 1pt solid windowtext;border-image: initial;border-top: none;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eStumble\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: red;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e22%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.7pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(0, 176, 80);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e78%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.8pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.15pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: red;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e4%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.8pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(0, 176, 80);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e92%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 48.05pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: red;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e4%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 62.75pt;border-right: 1pt solid windowtext;border-bottom: 1pt solid windowtext;border-left: 1pt solid windowtext;border-image: initial;border-top: none;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eWalk\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.7pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: red;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e30%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.8pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(0, 176, 80);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e70%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.15pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.8pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 48.05pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(0, 176, 80);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e100%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 62.75pt;border-right: 1pt solid windowtext;border-bottom: 1pt solid windowtext;border-left: 1pt solid windowtext;border-image: initial;border-top: none;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" style=\"width: 194.2pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eFallCNN_3\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" style=\"width: 193.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eFallCNN_4\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 62.75pt;border-right: 1pt solid windowtext;border-bottom: 1pt solid windowtext;border-left: 1pt solid windowtext;border-image: initial;border-top: none;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eFall\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(0, 176, 80);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e100%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.7pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.8pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.15pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(0, 176, 80);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e99%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.8pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: red;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e1%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 48.05pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height: normal;font-size:15px;font-family:\"Calibri\",sans-serif;'\u003e0%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 62.75pt;border-right: 1pt solid windowtext;border-bottom: 1pt solid windowtext;border-left: 1pt solid windowtext;border-image: initial;border-top: none;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eJump\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: red;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e13%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.7pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(0, 176, 80);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e87%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.8pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.15pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: red;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e9%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.8pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(0, 176, 80);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e91%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 48.05pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 62.75pt;border-right: 1pt solid windowtext;border-bottom: 1pt solid windowtext;border-left: 1pt solid windowtext;border-image: initial;border-top: none;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eStumble\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: red;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e4%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.7pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(0, 176, 80);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e92%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.8pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: red;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e4%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.15pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: red;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e4%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.8pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(0, 176, 80);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e92%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 48.05pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: red;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e4%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 62.75pt;border-right: 1pt solid windowtext;border-bottom: 1pt solid windowtext;border-left: 1pt solid windowtext;border-image: initial;border-top: none;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eWalk\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.7pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: red;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e1%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.8pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(0, 176, 80);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e99%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.15pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.8pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 48.05pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(0, 176, 80);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e100%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:8.0pt;margin-left:0in;line-height:107%;font-size:15px;font-family:\"Calibri\",sans-serif;'\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:8.0pt;margin-left:0in;line-height:107%;font-size:15px;font-family:\"Calibri\",sans-serif;'\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:8.0pt;margin-left:0in;line-height:107%;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:justify;'\u003e\u003cspan style='font-size:16px;line-height:107%;font-family:\"Times New Roman\",serif;'\u003eTable 3.\u0026nbsp;\u003c/span\u003e\u003cspan style='font-size:16px;line-height:107%;font-family:\"Times New Roman\",serif;'\u003e\u0026nbsp;Performance Summary for FallCNN models\u003c/span\u003e\u003c/p\u003e\n \u003ctable style=\"border-collapse:collapse;border:none;\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 63.6pt;border: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" style=\"width: 193.55pt;border-top: 1pt solid windowtext;border-right: 1pt solid windowtext;border-bottom: 1pt solid windowtext;border-image: initial;border-left: none;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eFallCNN_1\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" style=\"width: 193.65pt;border-top: 1pt solid windowtext;border-right: 1pt solid windowtext;border-bottom: 1pt solid windowtext;border-image: initial;border-left: none;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eFallCNN_2\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 47.05pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eFall\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.8pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eJump\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eStumble\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eWalk\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.1pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eFall\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eJump\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eStumble\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eWalk\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 63.6pt;border-right: 1pt solid windowtext;border-bottom: 1pt solid windowtext;border-left: 1pt solid windowtext;border-image: initial;border-top: none;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eRecall\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.05pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.58\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.8pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e1\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.71\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e1\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.1pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.88\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.96\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.98\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.96\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 63.6pt;border-right: 1pt solid windowtext;border-bottom: 1pt solid windowtext;border-left: 1pt solid windowtext;border-image: initial;border-top: none;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003ePrecision\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.05pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.99\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.8pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.52\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.78\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.7\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.1pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.96\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.91\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.92\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e1\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 63.6pt;border-right: 1pt solid windowtext;border-bottom: 1pt solid windowtext;border-left: 1pt solid windowtext;border-image: initial;border-top: none;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eF1 Score\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.05pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.73\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.8pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.68\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.74\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.82\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.1pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.91\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.93\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.94\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.97\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 63.6pt;border-right: 1pt solid windowtext;border-bottom: 1pt solid windowtext;border-left: 1pt solid windowtext;border-image: initial;border-top: none;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eLoss\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" style=\"width: 193.55pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.264\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" style=\"width: 193.65pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.118\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 63.6pt;border-right: 1pt solid windowtext;border-bottom: 1pt solid windowtext;border-left: 1pt solid windowtext;border-image: initial;border-top: none;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eAccuracy\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" style=\"width: 193.55pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e94%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" style=\"width: 193.65pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e95%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 63.6pt;border-right: 1pt solid windowtext;border-bottom: 1pt solid windowtext;border-left: 1pt solid windowtext;border-image: initial;border-top: none;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e\u0026nbsp;\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" style=\"width: 193.55pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eFallCNN_3\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" style=\"width: 193.65pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eFallCNN_4\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 63.6pt;border-right: 1pt solid windowtext;border-bottom: 1pt solid windowtext;border-left: 1pt solid windowtext;border-image: initial;border-top: none;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eRecall\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.05pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.85\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.8pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e1\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.98\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.96\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.1pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.88\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.98\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e1\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.96\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 63.6pt;border-right: 1pt solid windowtext;border-bottom: 1pt solid windowtext;border-left: 1pt solid windowtext;border-image: initial;border-top: none;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003ePrecision\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.05pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e1\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.8pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.87\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.92\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.99\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.1pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.99\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.91\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.92\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e1\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 63.6pt;border-right: 1pt solid windowtext;border-bottom: 1pt solid windowtext;border-left: 1pt solid windowtext;border-image: initial;border-top: none;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eF1 Score\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.05pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.91\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.8pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.93\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.94\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.97\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.1pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.93\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.94\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 50.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.95\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 47.85pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.97\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 63.6pt;border-right: 1pt solid windowtext;border-bottom: 1pt solid windowtext;border-left: 1pt solid windowtext;border-image: initial;border-top: none;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eLoss\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" style=\"width: 193.55pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.098\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" style=\"width: 193.65pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e0.083\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 63.6pt;border-right: 1pt solid windowtext;border-bottom: 1pt solid windowtext;border-left: 1pt solid windowtext;border-image: initial;border-top: none;background: rgb(166, 166, 166);padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align: justify;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003eAccuracy\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" style=\"width: 193.55pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e97%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" style=\"width: 193.65pt;border-top: none;border-left: none;border-bottom: 1pt solid windowtext;border-right: 1pt solid windowtext;padding: 0in 5.4pt;vertical-align: top;\"\u003e\n \u003cp style='margin-top:0in;margin-right:0in;margin-bottom:.0001pt;margin-left:0in;line-height:normal;font-size:15px;font-family:\"Calibri\",sans-serif;text-align:center;'\u003e\u003cspan style='font-size:16px;font-family:\"Times New Roman\",serif;'\u003e98%\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cdiv class=\"Section3\" id=\"Sec12\"\u003e\n \u003ch2\u003e4.2.2. Transfer Learning Models\u003c/h2\u003e\n \u003cp\u003eThe result of performance metrics such as accuracy, confusion matrix, loss, recall, precision, and f1-score on the transfer learning models are shown in Table \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e and Table \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003e.\u003c/p\u003e\n \u003cp\u003eTable 4.\u0026nbsp;\u0026nbsp;Confusion matrix for\u0026nbsp;MobileNetV2\u0026nbsp;models\u003c/p\u003e\n \u003ctable border=\"1\" cellpadding=\"0\" cellspacing=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" valign=\"top\" width=\"13.96103896103896%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" valign=\"top\" width=\"43.01948051948052%\"\u003e\n \u003cp\u003e\u003cstrong\u003eMobileNetV2 0.05\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" valign=\"top\" width=\"43.01948051948052%\"\u003e\n \u003cp\u003e\u003cstrong\u003eMobileNetV2 0.1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"12.406015037593985%\"\u003e\n \u003cp\u003eFall\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"12.406015037593985%\"\u003e\n \u003cp\u003eJump\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"12.781954887218046%\"\u003e\n \u003cp\u003eStumble\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"12.406015037593985%\"\u003e\n \u003cp\u003eWalk\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"12.406015037593985%\"\u003e\n \u003cp\u003eFall\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"12.406015037593985%\"\u003e\n \u003cp\u003eJump\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"12.781954887218046%\"\u003e\n \u003cp\u003eStumble\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"12.406015037593985%\"\u003e\n \u003cp\u003eWalk\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"13.915857605177994%\"\u003e\n \u003cp\u003eFall\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e100%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"11.003236245954692%\"\u003e\n \u003cp\u003e0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e99%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e1%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"11.003236245954692%\"\u003e\n \u003cp\u003e0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e0%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"13.915857605177994%\"\u003e\n \u003cp\u003eJump\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e34%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e66%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"11.003236245954692%\"\u003e\n \u003cp\u003e0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e14%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e86%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"11.003236245954692%\"\u003e\n \u003cp\u003e0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e0%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"13.915857605177994%\"\u003e\n \u003cp\u003eStumble\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e6%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"11.003236245954692%\"\u003e\n \u003cp\u003e94%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"11.003236245954692%\"\u003e\n \u003cp\u003e92%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e8%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"13.915857605177994%\"\u003e\n \u003cp\u003eWalk\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"11.003236245954692%\"\u003e\n \u003cp\u003e0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e100%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"11.003236245954692%\"\u003e\n \u003cp\u003e17%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e83%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eTable 5.\u0026nbsp;\u0026nbsp;Performance summary for\u0026nbsp;MobileNetV2\u0026nbsp;models\u003c/p\u003e\n \u003ctable border=\"1\" cellpadding=\"0\" cellspacing=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" valign=\"top\" width=\"13.96103896103896%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" valign=\"top\" width=\"43.01948051948052%\"\u003e\n \u003cp\u003eMobileNetV2 0.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" valign=\"top\" width=\"43.01948051948052%\"\u003e\n \u003cp\u003eMobileNetV2 0.1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"12.406015037593985%\"\u003e\n \u003cp\u003eFall\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"12.406015037593985%\"\u003e\n \u003cp\u003eJump\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"12.781954887218046%\"\u003e\n \u003cp\u003eStumble\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"12.406015037593985%\"\u003e\n \u003cp\u003eWalk\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"12.406015037593985%\"\u003e\n \u003cp\u003eFall\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"12.406015037593985%\"\u003e\n \u003cp\u003eJump\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"12.781954887218046%\"\u003e\n \u003cp\u003eStumble\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"12.406015037593985%\"\u003e\n \u003cp\u003eWalk\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"13.915857605177994%\"\u003e\n \u003cp\u003eRecall\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e0.71\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"11.003236245954692%\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e0.87\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e0.98\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"11.003236245954692%\"\u003e\n \u003cp\u003e0.84\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e0.91\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"13.915857605177994%\"\u003e\n \u003cp\u003ePrecision\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e0.66\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"11.003236245954692%\"\u003e\n \u003cp\u003e0.94\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e0.86\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"11.003236245954692%\"\u003e\n \u003cp\u003e0.92\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e0.83\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"13.915857605177994%\"\u003e\n \u003cp\u003eF1 Score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e0.83\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e0.79\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"11.003236245954692%\"\u003e\n \u003cp\u003e0.96\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e0.92\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e0.91\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"11.003236245954692%\"\u003e\n \u003cp\u003e0.87\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"10.679611650485437%\"\u003e\n \u003cp\u003e0.86\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"13.96103896103896%\"\u003e\n \u003cp\u003eLoss\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" valign=\"top\" width=\"43.01948051948052%\"\u003e\n \u003cp\u003e0.19\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" valign=\"top\" width=\"43.01948051948052%\"\u003e\n \u003cp\u003e0.07\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"13.96103896103896%\"\u003e\n \u003cp\u003eAccuracy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" valign=\"top\" width=\"43.01948051948052%\"\u003e\n \u003cp\u003e97.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" valign=\"top\" width=\"43.01948051948052%\"\u003e\n \u003cp\u003e97.2%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cdiv class=\"Section3\" id=\"Sec13\"\u003e\n \u003ch2\u003e4.2.3 Overall performance\u003c/h2\u003e\n \u003cp\u003eTable \u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003e shows the overall validation accuracy (Accuracy, F1 Score, Recall, Precision) for the custom FallCNN models and the transfer learning models.\u003c/p\u003e\n \u003c/div\u003e\n\u003c/div\u003e\n\u003cdiv class=\"Section2\" id=\"Sec14\"\u003e\n \u003ch2\u003e4.3. Feature Explorer\u003c/h2\u003e\n \u003cp\u003eThe feature explorers for the transfer learning models MobileNetV2 0.05 and MobileNetV2 0.1 on the training data generated with the help of edge impulse platform are shown in Figs. \u003cspan class=\"InternalRef\"\u003e9\u003c/span\u003e and \u003cspan class=\"InternalRef\"\u003e10\u003c/span\u003e.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv class=\"Section2\" id=\"Sec15\"\u003e\n \u003ch2\u003e4.4. Visual Interpretation of The Trained Model Features\u003c/h2\u003e\n \u003cp\u003eFigure \u003cspan class=\"InternalRef\"\u003e11\u003c/span\u003e shows the visualization of features extracted by the inner layers of the proposed FallCNN_4 during the training process. This visualization helps a lot in understanding how a model is interpreting the image internally. The salient features on the convolution and pooling layer of FallCNN_4 is visualized since this model achieved better results compared to other versions.\u003c/p\u003e\n\u003c/div\u003e"},{"header":"5. Conclusion","content":"\u003cp\u003eElderly Human Activity Recognition using computer vision has attained huge research attention. Recognizing the elderly movement like fall, Jump Stumble and Walk with the help of AI based intelligent analysis can be benefited for elderly monitoring. FDEP analysed various elderly fall detection on IoT-based accelerometer/gyroscope devices. To make FEDP more effective, a deep learning based custom FallCNN models are designed for enhanced eldely monitoring. A SimgFall graph images dataset are created by plotting the sensor data obtained from the SiSFall dataset for intelligent prediction of the selected activity such as fall, jump, stumble, and walk.\u003c/p\u003e \u003cp\u003eThe new dataset was tested with six models (four custom FallCNN models and two transfer learning models like MobileNetV2 0.05, MobileNetV2 0.1). In the case of transfer learning models, MobileNetV2 0.05 performed well when compared to the MobileNetV2 0.1. Similarly, FallCNN_4 provided better results among all the custom FallCNN models and the transfer learning models adopted in the proposed work. The performance metrics of the FallCNN_4 model with an accuracy of 98% outperforming all other Custom CNN (FallCNN) and Transfer learning models (MobileNetV2).\u003c/p\u003e \u003cp\u003eOur future work includes\u003c/p\u003e \u003cp\u003e(a) Working on real time elderly data collection more aligned with real-life scenarios, with more activities to analyse and predict.\u003c/p\u003e \u003cp\u003e(b) Abnormal elderly behaviour prediction to evaluate their activity and to support health care services.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;Not applicable.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eM. Vergin Raja Sarobin and \u0026nbsp; L. Jani Anbarasi formulated the work and M. Vergin Raja Sarobin and Jishnuraj k completed the implementation and M. Vergin Raja Sarobin , L. Jani Anbarasi and Jishnuraj. K wrote the main manuscript text , Rukmani P, S Graceline Jasmine and Modigari Narendra prepared the figures and validated the results. All authors reviewed the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAll data generated or analysed during this study are included in this published article.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;Not applicable.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eNot applicable.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003e\u003cem\u003eWorld Health Organization. WHO Global Report on Falls Prevention in Older Age; World Health Organization: Geneva, Switzerland, 2007.\u003c/em\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eS. Lord, S. Smith, \u003cem\u003eand J. Menant, \u0026ldquo;Vision and falls in older people: Risk factors and intervention strategies,\u0026rdquo; Clin. Geriatric Med., vol. 26, pp. 569\u0026ndash;581\u003c/em\u003e, 2010.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIgual, R.; Medrano, C.; Plaza, I. \u003cem\u003eChallenges, issues and trends in fall detection systems. BioMedical Engineering OnLine\u003c/em\u003e 2013, \u003cem\u003e12, 1\u0026ndash;24.\u003c/em\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eM.Mubashir, L. Shao, \u003cem\u003eand\u003c/em\u003e L. Seed, \u003cem\u003e\u0026ldquo;A survey on fall detection: Principles and approaches,\u0026rdquo; Neurocomputing, vol. 100, pp. 144\u0026ndash;152\u003c/em\u003e, 2013.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eL. Montesinos, R. Castaldo, \u003cem\u003eand\u003c/em\u003e L. Pecchia, \u003cem\u003e\u0026ldquo;Wearable inertial sensors for fall risk assessment and prediction in older adults: A systematic review and meta-analysis,\u0026rdquo; IEEE Trans. Neural Syst. Rehabil. Eng., vol. 26, no. 3, pp. 573\u0026ndash;582, Mar.\u003c/em\u003e 2018.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eN. Pannurat, S. Thiemjarus, \u003cem\u003eand E. Nantajeewarawat, \u0026ldquo;Automatic fall monitoring: A review,\u0026rdquo; Sensors, vol. 14, pp. 12900\u0026ndash;12936\u003c/em\u003e, 2014.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRimminen H., Lindstr\u0026ouml;m J., Linnavuo M., Sepponen \u003cem\u003eR. Detection of falls among the elderly by a floor sensor using the electric near field\u003c/em\u003e. \u003cem\u003eIEEE Trans. Inf. Technol. Biomed.\u003c/em\u003e 2010;\u003cem\u003e14\u003c/em\u003e:\u003cem\u003e1475\u0026ndash;1476\u003c/em\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGjoreski H., Lustrek M., Gams M. \u003cem\u003eAccelerometer placement for posture recognition and fall detection. Proceedings of the Seventh International Conference on Intelligent Environments; Nottingham, UK. 25\u0026ndash;28 July\u003c/em\u003e 2011; \u003cem\u003epp. 47\u0026ndash;54. [Google Scholar]\u003c/em\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi Q., Stankovic J.A. \u003cem\u003eGrammar-based, posture- and context-cognitive detection for falls with different activity levels. Proceedings of the Second Conference on Wireless Health; La Jolla, CA, USA. 10\u0026ndash;13 October\u003c/em\u003e 2011; \u003cem\u003epp. 1\u0026ndash;10.\u003c/em\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eE. E. Stone \u003cem\u003eand M. Skubic, \u0026ldquo;Fall detection in homes of older adults using the Microsoft Kinect,\u0026rdquo; IEEE J. Biomed. Health Informat., vol. 19, no. 1, pp. 290\u0026ndash;301, Jan.\u003c/em\u003e 2015.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTzeng H.-W., Chen M.-Y., Chen J.-Y. \u003cem\u003eDesign of fall detection system with floor pressure and infrared image. Proceedings of the International Conference on System Science and Engineering; Taipei, Taiwan. 1\u0026ndash;3 July\u003c/em\u003e 2010; \u003cem\u003epp. 131\u0026ndash;135. [Google Scholar]\u003c/em\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVavoulas, G.; Pediaditis, M.; Chatzaki, C.; Spanakis, E.; Tsiknakis,\u003cem\u003eM. The MobiFall Dataset: Fall Detection and Classification with a Smartphone. International Journal of Monitoring and Surveillance Technologies Research\u003c/em\u003e 2014, \u003cem\u003e2, 44\u0026ndash;56\u003c/em\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMedrano, C.; Igual, R.; Plaza, I.; Castro, M. \u003cem\u003eDetecting Falls as Novelties in Acceleration Patterns Acquired with Smartphones\u003c/em\u003e. \u003cem\u003ePlos One\u003c/em\u003e 2014, \u003cem\u003e9\u003c/em\u003e, \u003cem\u003ee94811\u003c/em\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSucerquia A, L\u0026oacute;pez JD, Vargas-Bonilla JF. \u003cem\u003eSisFall: A Fall and Movement Dataset. Sensors.\u003c/em\u003e 2017;\u003cem\u003e17(1).\u003c/em\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eL. J. Kau and C. S. Chen, \u003cem\u003e\u0026ldquo;A smart phone-based pocket fall accident detection, positioning, and rescue system,\u0026rdquo; IEEE J. Biomed. Health Informat., vol. 19, no. 1, pp. 44\u0026ndash;56, Jan.\u003c/em\u003e 2015.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eG. Sannino, I. De Falco, \u003cem\u003eand G. De Pietro, \u0026ldquo;A supervised approach to automatically extract a set of rules to support fall detection in an mHealth system,\u0026rdquo; Appl. Soft Comput., vol. 34, pp. 205\u0026ndash;216\u003c/em\u003e, 2015.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePalmerini, L., Klenk, J., Becker, C. and Chiari, L., 2020. \u003cem\u003eAccelerometer-based fall detection using machine learning: training and testing on real-world falls. Sensors, 20(22), p.6479.\u003c/em\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e\u003cem\u003eJ. Silva et al., \u0026ldquo;Comparing machine learning approaches for fall risk assessment,\u0026rdquo; in Proc. 10th Int. Joint Conf. Biomed. Eng. Syst. Technol.\u003c/em\u003e, 2017, \u003cem\u003epp. 223\u0026ndash;230.\u003c/em\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJ. Howcroft, E. D. Lemaire, \u003cem\u003eand J. Kofman, \u0026ldquo;Prospective elderly fall prediction by older-adult fall-risk modeling with feature selection,\u0026rdquo; Biomed. Signal Process. Control, vol. 43, pp. 320\u0026ndash;328\u003c/em\u003e, 2018.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRodrigues, T.B., Salgado, D.P., Cordeiro, M.C., Osterwald, K.M., Teodiano Filho, F.B., \u003cem\u003ede Lucena Jr\u003c/em\u003e, V.F., Naves, \u003cem\u003eE.L. and\u003c/em\u003e Murray, N., 2018. \u003cem\u003eFall detection system by machine learning framework for public health. Procedia Computer Science, 141, pp.358\u0026ndash;365.\u003c/em\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eY. LeCun, Y. Bengio, \u003cem\u003eand\u003c/em\u003e G. Hinton, \u003cem\u003e\u0026ldquo;Deep learning,\u0026rdquo; Nature, vol. 521, pp. 436\u0026ndash;444\u003c/em\u003e, 2015.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eA. Nunez-Marcos, \u003cem\u003eC.Azkune, and I. Arganda-Carreras, \u0026ldquo;Vision-based fall detection with convolutional neural networks,\u0026rdquo; Wireless Commun. Mobile Comput., vol.\u003c/em\u003e 2017, \u003cem\u003e2017, Art. no. 9474806.\u003c/em\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLu, N., Wu, Y., Feng, L. and Song, J., 2018. Deep learning for fall detection: Three-dimensional CNN combined with LSTM on video kinematic data. IEEE journal of biomedical and health informatics, 23(1), pp.314\u0026ndash;323.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003eTable 1 and 6 are available in the Supplementary Files section.\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"CNN, Transfer learning model, Fall detection, Deep learning, Performance analysis","lastPublishedDoi":"10.21203/rs.3.rs-1884093/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-1884093/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eAs per the data provided by the World Health Organization (WHO), falls are one of the major reasons for unintentional deaths or injuries in elderly people. Even though there are a lot of fall detection methods and algorithms exist, there is no efficient artificial intelligence strategy for detecting falls. Various literature states that Fall Detection among Elderly Person (FDEP) provides the possibility of bringing up an efficient and cost-effective way to tackle this problem. This paper generated a signal-based image dataset, SimgFall from the existing accelerometer or gyroscope-based sensor data of the SiSFall dataset for early detection of fall to fasten the medical assistance process. This SimgFall data is proposed by the proposed FallCNN model, a novel deep Convolutional Neural Network (CNN) based architecture that includes multiple folds of CNN network. These models utilize depth-wise convolution with varying dilation rates for efficiently extracting diversified features from the SimgFall dataset. 1992 signal-based images of which 498 are the samples collected for fall, jump, stumble and walk of four classes respectively. Further, performance evaluation on the generated dataset using different pre-trained and custom models has been analysed based on the loss and accuracy curve. The experimental results show that the highest classification accuracy (98%) is achieved using the proposed customized FallCNN model.\u003c/p\u003e","manuscriptTitle":"Fall Detection among Elderly Person using FallCNN and Transfer Learning Models","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2022-08-01 17:36:48","doi":"10.21203/rs.3.rs-1884093/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"e0954247-390c-4f98-9ff2-15b9efa6d92b","owner":[],"postedDate":"August 1st, 2022","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2022-12-14T02:29:16+00:00","versionOfRecord":[],"versionCreatedAt":"2022-08-01 17:36:48","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-1884093","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-1884093","identity":"rs-1884093","version":["v1"]},"buildId":"7rjqhiLT3MXkJMwkYKINL","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}