Application of Machine Learning in predicting intimate partner-based violence among reproductive age women in Rwanda

Application of Machine Learning in predicting intimate partner-based violence among reproductive age women in Rwanda | 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 Application of Machine Learning in predicting intimate partner-based violence among reproductive age women in Rwanda Musa Nget, Aliane Uwera, Eva Moeng, Amadou Barrow, Jam Camara, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7686865/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 7 You are reading this latest preprint version Abstract Background Intimate partner violence (IPV) is a serious global public health issue that predominantly affects women. Despite its widespread nature, IPV data and prevention strategies remain limited, particularly in low- and middle-income countries (LMICs), with sub-Saharan Africa bearing a significant burden. This study seeks to develop and validate a robust machine learning model to predict IPV and its predictors in Rwanda. Methods The study used five different machine learning algorithms, including the Least Absolute Shrinkage and Selection Operator (LASSO), logistic regression (LR), Random Forest, Support Vector Machine, and Classification and Regression Trees (CART) to develop models using data of 14,634 participants from the 2019–2020 Demographic and Health Survey. We used multiple imputation for missing data, and the Shapley Additive Explanations (SHAP) technique was used to select the most predictive variables. The data was randomly split into 80% and 20% for training and testing, respectively. To achieve a balanced dataset, the Synthetic Minority Over-sampling Technique was used on the training subset. The algorithms were evaluated using F1 score, AUC, accuracy, precision, sensitivity, and specificity. Results The random forest algorithm (accuracy and Receiver Operating Characteristic (ROC) curve of 99.1% and 100%, respectively) proved remarkably successful in accurately predicting IPV, outperforming other algorithms. Interestingly, the model’s performance revealed a significant improvement when utilizing only fifteen variables to predict IPV. The algorithm owes its success to its remarkable capability to identify crucial predictor features, with the top four being arguing with a partner, jealous partner, burning food, and insisting on knowing the respondent’s whereabouts. Conclusion The random forest algorithm was a highly effective tool for identifying potential predictors of IPV. Machine learning models may serve as an effective approach to address IPV and could help in early intervention and prevention by strengthening policy enforcement and integrating predictive tools into healthcare systems. Machine Learning Random Forest Algorithm Intimate Partner Violence Predictive Model Rwanda Figures Figure 1 Figure 2 Figure 3 Introduction Intimate partner violence (IPV) is a serious global public health issue that predominantly affects women, although men can also be victims 1 . According to the World Health Organization (WHO), IPV is defined as acts by a current or former intimate partner that cause physical, sexual, or psychological harm, including physical aggression, sexual coercion, psychological abuse, and controlling behaviours 1 . The Center for Disease Control and Prevention (CDC) typically categorizes IPV into four types: sexual violence, physical violence, threats, and psychological or emotional abuse 2 . Globally, the prevalence of IPV remains high, especially in low-income countries. Approximately 27% of women aged 15–49 in intimate relationships report experiencing IPV, and IPV accounts for 38% of female homicides worldwide 1 . Despite its widespread nature, IPV data and prevention strategies remain limited, particularly in low- and middle-income countries (LMICs), with sub-Saharan Africa bearing a significant burden 3 . Previous research has shown that women who have experienced physical or sexual intimate partner violence are at risk of developing physical problems 4 , 5 , as well as mental health issues such as depression, anxiety, posttraumatic stress disorder, suicide, and being drunkards 6 . The Global Burden of Disease highlights that the highest prevalence of IPV occurs in central sub-Saharan Africa (32%) and Oceania (29%), followed by eastern sub-Saharan Africa (24%), and South Asia (19%) 7 . In Rwanda, the prevalence of IPV was reported to be 40.21% 8 , with a higher percentage of women (49.6%) seeking help 7 . The causes and effects of IPV on women, especially in Africa, are subject to various speculations and cultural norms. Rwanda is often regarded as a leader in Africa in promoting gender equality, mainly due to its progressive policies and strong legal frameworks that target gender-based violence, including IPV 9 . Like other countries around the world, Rwanda has rules and policies to protect women from violence, unlike other countries in the region. For example, in 2008, the Rwandan government passed the Prevention and Punishment of Gender-based Violence Law, which encompasses all types of violence and has a minimum prison penalty of six months. These initiatives have contributed to significant reductions in IPV and improvements in women's rights 10 . The previous studies conducted across different regions highlighted age, education level, alcohol and substance use, childhood exposure to violence, the duration of relationships, and partner’s controlling behaviour as the common emerging factors associated with IPV. These factors vary widely across cultural and socioeconomic settings. A recent study revealed that intimate partners with a certain level of education and those who had completed primary or higher education were more likely to report IPV 11 , 12 . Additionally, a cross-sectional study reported that women whose partners consumed alcohol reported IPV 2.3 times higher than their counterparts 13 . Gubi et al found that partners with longer periods of relationship had higher odds of experiencing IPV 14 . Moreover, growing up in a family that normalized violence was associated with IPV, whereby people who were exposed to violence during their childhood are prone to acceptance of IPV and are immersed in the mire of recurrent IPV 13 . Furthermore, various studies found that women in relationships with men who had controlling behaviours or had been involved in any form of aggressive behaviour were highly likely to experience IPV 14 , 15 . Although efforts to reduce intimate partner violence (IPV) have been put in place by different organizations and countries, it continues to pose a significant public health concern that necessitates more targeted interventions 8 . Despite IPV-related injuries continuing to be a leading cause of injury and death worldwide, data to inform prevention strategies remain limited in low and middle-income countries (LMICs), especially in sub-Saharan Africa 3 . Machine learning is a subfield of artificial intelligence that focuses on creating systems that learn patterns from data rather than being explicitly programmed with rules. It allows models to automatically learn from historical data to make accurate forecasts or classifications about future or unseen events 16 . Over the past decade, machine learning (ML) has emerged as a robust and promising research methodology in healthcare research throughout the early detection, diagnosis, treatment, and prognosis evaluation 17 . Researchers have applied ML models to study the risk factors for aggressive episodes of various types of violence. However, the development of ML algorithms is still in its infancy 18 . Thus, more research and investments are required, preferably in large and prospective groups, to boost the application of ML models in health care and clinical practice 19 . Therefore, this study seeks to develop and validate a robust machine learning model to predict IPV and its predictors in Rwanda, thus enabling timely interventions to prevent intimate partner violence. Method Study design and data source The study used data from the Rwanda Demographic and Health Survey 2019–2020, conducted by the National Institute of Statistics of Rwanda (NISR) in partnership with the Global Demographic and Health Surveys Program. A two-stage stratified cluster sample design was used to collect data from 14,634 women within the reproductive age group of 15–49 years in Rwanda. Initially, clusters were selected from enumeration areas, followed by the selection of households within these clusters. RDHS 2019–2020 was a cross-sectional, nationally representative survey comprising multiple datasets, including but not limited to the IPV dataset. This study used the individual file record (RWFW81), which includes data on intimate partner violence. Outcome variable Ever-witnessed IPV was the dependent variable in the present study. Two categories, “Yes” (experienced IPV) and “No” (did not experience IPV), were used to classify the dependent variable. One indicates “Yes”, whereas zero indicates “No”. Physical violence (pushing, slapping, punching with a fist or something harmful, kicking or beating, threats, or strangulation or burning). Emotional violence (humiliation in front of others, insults and negative self-perception, intimidation, isolation from friends and family, jealousy and accusations of infidelity, and control over movements and activities). Sexual violence (forced sexual acts against her will, physical injuries resulting from sexual violence, forced sexual intercourse during pregnancy, experiences involving multiple perpetrators, or forced sexual acts perceived as humiliation). Features This study explored multiple factors that influence intimate partner violence through three key categories. Sociodemographic features included residence type, age, education, marital status, region, religion, wealth, travel pattern, and cohabitation status. Behavioural features examined included relationship dynamics, sexual autonomy, alcohol use, communication monitoring, and reproductive choices, including forced sexual initiation and abortion history. The psychosocial features focused on decision-making autonomy regarding finances, healthcare, social interactions, and childhood exposure to violence. Healthcare and structural vulnerability factors specifically address physical harm during pregnancy caused by the former or previous partner. Data pre-processing This study involved various steps to prepare the RWDHS 2019–2020 dataset. The dataset consisted of 60 features with a single target feature. The data underwent a cleaning process, and missing values were handled using the K-nearest neighbor approach and multiple imputation. Initially, to prepare the data for the model, it is necessary to convert the input and output features into numerical values. The researchers used a one-hot encoder to encode the categorical variables found in the dataset. This technique transforms each categorical value into a new column with one-hot encoding, and the label values are created as new columns with values of either 1 or 0. Secondly, we used multiple imputation techniques to address missing values and incomplete raw data. As detailed in the supplementary figures, a breakdown of the percentage of missing values for each feature is as follows: argues with partner 4582 (31.3%), jealous partner 291(2%), burns food 2632(18%), insists on knowing respondent’s whereabouts 737(5%), refuses to have sex 4550 (31.1%), partner accusation of unfaithfulness 291(2%), rural residence 11,083(75.7%), alcohol consumption 1230(8.4%), residing 6511 (44.5%), wife looking at the phone (missing), child embarrasses family (missing), and sex outside marriage (missing). Subsequently, duplicates were screened and eliminated from the final dataset. Feature selection SHAP (Shapley Additive Explanations) is a technique for elucidating machine learning model predictions by calculating feature contributions derived from Shapley values in cooperative game theory 20 . It assigns the disparity between a model's forecast and baseline value to each feature, guaranteeing coherence and local precision in the explanation. Imbalanced data handling Imbalanced data in machine learning poses a significant problem because it can result in bias that favors the majority class. This difficulty typically arises when the distribution of classes within a dataset is unbalanced. To achieve dataset balance in the current study, we used the Synthetic Minority Over-sampling Technique (SMOTE). SMOTE is an oversampling technique that generates synthetic instances of the minority class through the computation of existing minority samples. This method balances class distribution in the training dataset, enhancing the classifier's capacity to learn from imbalanced data 21 . We evaluated its effectiveness by comparing the chosen supervised machine learning algorithms, emphasizing the accuracy and AUC. This study illustrates that the use of balanced sampling techniques is essential. Model development This study used machine learning algorithms, including logistic regression (LR), Classification and Regression Trees (CART), Least Absolute Shrinkage and Selection Operator (LASSO), Support Vector Machine (SVM), and Random Forest (RF) to predict intimate partner violence. Logistic regression (LR) is one of the earliest and most commonly used classification algorithms in machine learning that predicts the probability of a binary outcome based on one or more input variables by modeling the log-odds of the outcome as a linear combination of the input features 16 , 22 . LR was used as a foundational model for selecting input features in the prediction model, then feature selection was performed using the Shapley additive explanation (SHAP) method to identify the most relevant predictors. Subsequently, we used four ML algorithms (CART, LASSO, SVM, RF) to evaluate their effectiveness in predicting violent behavior among intimate partners. Classification and Regression Trees is a decision tree algorithm for classification and regression tasks. It splits the dataset into subsets based on feature values that result in the most homogeneous subgroups 16 . Least Absolute Shrinkage and Selection Operator is a linear regression algorithm with regularization applied to high-dimensional analysis 23 . Support Vector Machine (SVM) is a supervised classification algorithm that aims to find the optimal decision boundary that maximizes the margin between classes in the data points by transforming input data into a high-dimensional feature space using kernel functions. This method allows the model to handle non-linear relationships 16 . SVM has a strong theoretical underpinning and has revealed an effective performance on various tasks 24 . Random forest (RF) is a supervised ensemble learning method that generates a single output or prediction by integrating the outcomes of various decision trees. It reduces overfitting through random sampling of both data points and features 16 . Model evaluation (Prediction Model Evaluation) The researchers used k-fold cross-validation (k = 10) to guarantee the strength and applicability of the machine learning model for predicting intimate partner violence (IPV) and to assess its performance. This approach reduces the risk of overfitting by dividing the dataset into k subsets, where the model is systematically trained on k-1 folds and validated on the left-out fold 25 . We divided the training set into ten equal subsets, one for validation and the remaining nine for training. Upon completion of cross-validation, the accuracy and kappa metrics were utilized to produce the final models for each algorithm. The evaluation of these models on the test set involved various performance metrics, such as F1 score, AUC, accuracy, precision, sensitivity, and specificity. The model with the highest AUC value was identified as the optimal choice. Figure 1 : Workflow of variable selection and model development. RF: Random forest, LR: Logistic regression, Lasso: least absolute shrinkage and selection operator, SVM: Support vector machine, CART: Classification and Regression Trees Results Participant’s characteristics Table 1 presents the characteristics of the study population after excluding participants who did not pertain to intimate partner violence. The study population comprised 14,634 individuals, with the majority residing in rural areas (75.7%) and a mean age of 29.1 (± 9.9) years. Nearly half (46.2%) were Protestants, followed by Catholics (37.6%). Most participants had partners with a primary education (58.1%), while 9.2% had no education. Marital status varied, with 41.4% never in union, 32.2% married, and 17.7% living with a partner. Geographically, participants were distributed across all regions, with the highest proportion in the East (24.8%). The wealth distribution was relatively even, with the poorest (19.4%) and richer (19.7%) categories representing almost similar proportions. Table 1 Sociodemographic characteristics of the study population Features Category Frequency (n) Percentage (%) Residence Rural 11,083 75.7 Urban 3,551 24.3 Age* 15–19 3,308 22.6 20–24 2,424 16.6 25–29 2,047 14.0 30–34 2,095 14.3 35–39 2,043 14.0 40–44 1,487 10.2 45–49 1,230 8.4 Partner’s education Higher 672 4.6 No education 1,352 9.2 Primary 8,500 58.1 Secondary 4,110 28.1 Marital status Divorced 379 2.6 Living with partner 2,584 17.7 Married 4,706 32.2 Never in union 6,060 41.4 Separated 527 3.6 Widowed 378 2.6 Region East 3,625 24.8 Kigali 1,921 13.1 North 2,294 15.7 South 3,482 23.8 West 3,312 22.6 Religion Adventist 1,842 12.6 Catholic 5,506 37.6 Muslim 287 2.0 No religion 114 0.8 Protestant 6,754 46.2 Traditional 1 0.0 Unknown 130 0.9 Wealth index Middle 2,709 18.5 Poorer 2,707 18.5 Poorest 2,844 19.4 Richer 2,884 19.7 Richest 3,490 23.8 Trips in the last 12 months 90+ 21 0.1 None 8,440 57.7 Unknown 6,173 42.2 Currently residing with partner Yes 6,511 44.5 No 8123 55.5 Intimate Partner Violence Experience IPV 149 1.02 No IPV 38 0.26 NA 14447 98.72 * Mean (standard deviation): 29.13 ± 9.92 Feature selection The analysis of feature importance across five machine learning models: Logistic Regression, Lasso, Random Forest, CART, and SVM, revealed key predictors associated with IPV (Fig. 1 , supplementary file Fig. S1 -S5). The most prominent predictor, which showed the highest importance score across almost all models, was arguing with a partner and was particularly influential in the logistic regression model (9.995), indicating that frequent disputes strongly correlate with the outcome. Other significant predictors in the logistic regression model included a jealous partner (2.794) and burning food (1.629), with jealousy showing notable importance in the Random Forest (0.044) and CART (0.109) models. Controlling behaviors such as insisting on knowing the respondent’s whereabouts (1.024) and accusations of unfaithfulness (0.627) were consistently influential across multiple models. Sexual behavior issues, such as refusing to have sex (0.645) and engaging in sex outside marriage (0.247), also contributed, albeit to a lesser extent. For other factors, residence (0.534) and alcohol consumption (0.403) had relatively low but non-negligible importance, with alcohol use appearing in several models. Methodologically, Logistic Regression and Random Forest models assigned greater weights to these features compared to the other models, suggesting differences in how these models prioritize predictors. Development and evaluation of the model The ML models demonstrated varying performance in predicting IPV, with tree-based methods (Random Forest and CART) outperforming other methods (Table 2 and Fig. 3 ). Random Forest achieved perfect discrimination (AUC = 1.000) and near-perfect accuracy (0.991), precision (0.995), recall (0.986), and specificity (0.995), indicating exceptional ability to identify both true positives and negatives. Similarly, CART exhibited strong performance (AUC = 0.986, accuracy = 0.986) with balanced precision (0.986) and recall (0.985). Logistic regression also performed well (AUC = 0.981, accuracy = 0.962), but had a slightly lower recall (0.928), suggesting minor under-detection of true IPV cases. In contrast, Lasso showed moderate discrimination (AUC = 0.821) but poor specificity (0.24), reflecting a high false-positive rate despite its high recall (0.97), whereas the SVM model performed inadequately (AUC = 0.569, recall = 0.293), failing to detect true IPV cases. These highlight Random Forest and CART as the most robust models for IPV prediction by combining high accuracy with balanced sensitivity and specificity. Logistic regression remains a viable alternative, while LASSO and SVM are unsuitable because of significant trade-offs in performance. Table 2 Evaluation of the model’s performance Models Accuracy Precision Recall Specificity F1 score Logistic Regression 0.962 0.995 0.928 0.995 0.960 Random Forest 0.991 0.995 0.986 0.995 0.991 Lasso 0.605 0.561 0.970 0.240 0.710 SVM 0.610 0.800 0.293 0.927 0.429 CART 0.986 0.986 0.985 0.986 0.986 Discussion Machine learning (ML) has emerged as a robust analysis method because of its ability to analyze large datasets and model complex relationships between risk factors and outcomes 26 . The current study represents a significant contribution to the application of machine learning for predicting IPV in Rwanda. Using socio-demographic, sociocultural, psychosocial, behavioural, sexual, healthcare, and structural features retrieved from the 2019–2020 Rwanda Demographic Health Survey, this study was able to predict the IPV status of individuals. After comparing five different ML algorithms, Random Forest emerged as the most effective algorithm for predicting IPV status. The research utilized SMOTE to overcome the dataset imbalance problem, thereby significantly improving the performance and reliability of the prediction model. To increase the interpretability of our predictions, we also performed a SHAP analysis to determine and measure each feature's contribution to our model's results. The model identified 15 significant factors among all variables considered in this study: arguing with partner, jealous partner, burning food, insistence on knowing the respondent's whereabouts, refusal to have sex, partner's accusations of unfaithfulness, place of residence, hurt during pregnancy by boyfriend, humiliation by previous partner, being physically forced to have sex by a previous partner, alcohol consumption, living with partner, wife looking through husband's phone, child embarrassing the family, and sex outside marriage. The results revealed that IPV was 1.02% among Rwandans. The findings contradict the global results, where a study conducted by the World Health Organization among 161 countries around the world found that 30% of women experience IPV in their lifetime 1 . Additionally, the current findings were lower than those of studies conducted in East African regions, 56% of Uganda 13 , 44% Tanzania 27, and also lower than those of studies conducted in other African regions (40% Gambia 11 , 42% Malawi 28 , 42.7% Zimbabwe 29 . The higher figures of IPV in African countries could be attributed to social and cultural acceptance of IPV as part of social interactions in relationships. Additionally, the difference in results could be attributed to the difference in research methodologies used among these studies; however, this implies that IPV remains a public issue, especially in African countries. Although some countries have implemented violence prevention policies, more effective policies and awareness are needed to reduce IPV. In the current study, most algorithms demonstrated high performance, particularly Random Forest, CART, and logistic regression, which showed strong and consistent results across all evaluation metrics. Random Forest achieved the highest accuracy and sensitivity scores of 99.1% and 98.6%, respectively. A previous study on violence among women similarly identified random forest as the best predictive model, with an accuracy and sensitivity score of 77% and 78%, respectively 30 . On the other hand, a study that used the Liberian 2019 DHS dataset and seven different machine learning algorithms found that decision tree and CatBoost outperformed the others in identifying women at risk of domestic violence, with an accuracy rate of 82%, while random forest revealed an accuracy rate of 81% 31 . This can be attributed to its ensemble learning approach, which lowers overfitting while increasing generalizability and handling imbalanced secondary data. Therefore, our findings emphasize the predictive power of Random for IPV. Hossain et al. revealed that machine learning algorithms, particularly random forests, successfully predicted domestic violence. The random forest method provided better results across all metrics 30 , similar to the results of the current study. With an AUC of 94%, Random Forest outperformed other algorithms in their research 31 . The random forest model was the best IPV predictive model in this study, with an AUC of 100%. The higher AUC of our model emphasizes that random forest accuracy predicts intimate partner violence by successfully distinguishing between positive and negative cases. Our study found several significant predictors of IPV; these features represent both interpersonal conflict and cultural standards that influence gender roles in justifying IPV. For instance, arguing with a partner, being jealous of a partner, burning food, refusal to have sex, and sex outside marriage justify the likelihood of experiencing IPV. Notably, a previous study emphasized that victims justify IPV under specific circumstances (refusal to have sex and partner infidelity) based on traditional acceptance 32 . This suggests that more women are at risk of experiencing IPV, as justification for such violence is frequently based on societal norms, tolerance, and those who may try to seek help are stigmatized and viewed as failing to be submissive in marriage. Strengths and limitations The strengths of this study lie in its use of a national dataset and advanced ML techniques to identify key predictors of IPV. The current research uses SMOTE, which helps balance the class distribution and improve the model's ability to generalize and predict minority classes accurately. Moreover, we employed SHAP to explain why RF exhibited the best performance among ML algorithms. SHAP solves the problem of multicollinearity by considering not only the influence of a single feature but also the synergy between features. However, significant improvements have been made in both the features and results, resulting in a more reliable model. This study has some limitations, including reliance on self-reported data, which may introduce bias, and the cross-sectional nature of the survey, which precludes causal inferences. Conclusion This study applied ML models to predict IPV among reproductive-aged women in Rwanda, using data from the 2019–2020 Rwanda Demographic and Health Survey. The Random Forest algorithm outperformed other models (logistic regression, CART, Lasso, SVM) with exceptional accuracy (99.1%) and sensitivity (98.6%), effectively identifying key predictors such as interpersonal conflicts (e.g., arguing with a partner, jealousy), controlling behaviours, and cultural justifications for violence (e.g., refusal to have sex, accusations of infidelity). Machine learning models that incorporate various features may serve as an effective approach to address IPV and could help in early intervention and prevention by strengthening policy enforcement, integrating predictive tools into healthcare systems, and launching community awareness campaigns to address the underlying cultural norms. Abbreviations ADASYN: Adaptive Synthetic Sampling AUC: Area Under the ROC Curve CART: Classification and Regression Trees CDC: The Center for Disease Control and Prevention IPV: Intimate partner violence LASSO: Least Absolute Shrinkage and Selection Operator LMICs: low- and middle-income countries LR: logistic regression ML: Machine learning NISR: National Institute of Statistics of Rwanda RF: Random forest ROC: Receiver Operating Characteristic RWDHS - Rwanda Demographic and Health Survey SHAP: Shapley Additive Explanations SMOTE: Synthetic Minority Oversampling Technique SVM: Support Vector Machine WHO: World Health Organization Declarations Data availability statement The data that support the findings of this study are openly available in The DHS Program at https://dhsprogram.com/data/available-datasets.cfm, reference country Rwanda. Ethics statement Ethics approval for this study was not required as we used secondary data that is publicly accessible. The survey, in its original format, obtained ethical approval from the Rwanda Research Ethics Committee (RWIR81FL.SAV). Source: Rwanda 2019/2020 Demographic Health Survey, October 2024 (https://www.dhsprogram.com/data/available-datasets.cfm). We secured authorization from the DHS Program to utilize the Rwanda DHS dataset in our research. The DHS program adheres to the protocol of informed consent and maintains the principles of anonymity and confidentiality under the regulations governing data utilization. Additional information concerning DHS data and ethical standards can be found at: http://goo.gl/ny8T6X. Author contributions MN conceptualised the study, MN, UA, EM and AB analysis of data, MN, UA and EM wrote the introduction, results and discussion, JC, AB and MT and critically reviewed the manuscript for its intellectual content. MN and UA had final responsibility to submit for publication. MN and UA accept full responsibility for the finished work and/or the conduct of the study, had access to the data and controlled the decision to publish. Funding The authors declare that no financial assistance was obtained for the research, authoring, and/or publication of this paper. Acknowledgements The authors express their gratitude to the MEASURE DHS project for providing free access to source data and for their help. Conflict of interest The authors affirm that there were no financial or business ties that could be interpreted as potential conflicts of interest during the research. References World Health Organization. Violence Against Women Prevalence Estimates 2018: Global, Regional and National Prevalence Estimates for Intimate Partner Violence Against Women and Global and Regional Prevalence Estimates for Non-Partner Sexual Violence Against Women. 1st ed. Geneva: World Health Organization; 2021. 1 p. Idoko P, Ogbe E, Jallow O, Ocheke A. Burden of intimate partner violence in The Gambia - a cross sectional study of pregnant women. Reprod Health. 2015 Dec;12(1):34. Blair KJ, De Virgilio M, Dissak-Delon FN, Dang LE, Christie SA, Carvalho M, et al. Associations between social determinants of health and interpersonal violence-related injury in Cameroon: a cross-sectional study. BMJ Glob Health. 2022 Jan;7(1):e007220. Cheung M, Leung P, Tsui V. Asian Male Domestic Violence Victims: Services Exclusive for Men. J Fam Violence. 2009 Oct;24(7):447–62. Kapiga S, Harvey S, Muhammad AK, Stöckl H, Mshana G, Hashim R, et al. Prevalence of intimate partner violence and abuse and associated factors among women enrolled into a cluster randomised trial in northwestern Tanzania. BMC Public Health. 2017 Dec;17(1):190. Curry SJ, Krist AH, Owens DK, Barry MJ, Caughey AB, Davidson KW, et al. Screening for Intimate Partner Violence, Elder Abuse, and Abuse of Vulnerable Adults: US Preventive Services Task Force Final Recommendation Statement. JAMA. 2018 Oct 23;320(16):1678. Ayebeng C, Dickson KS, Ameyaw EK, Adde KS, Paintsil JA, Yaya S. Influence of type of violence on women’s help-seeking behaviour: Evidence from 10 countries in sub-Saharan Africa. Pretorius D, editor. PLOS ONE. 2024 Mar 8;19(3):e0297308. Tessema ZT, Gebrie WM, Tesema GA, Alemneh TS, Teshale AB, Yeshaw Y, et al. Intimate partner violence and its associated factors among reproductive-age women in East Africa:-A generalized mixed effect robust poisson regression model. Paul P, editor. PLOS ONE. 2023 Aug 18;18(8):e0288917. Rwanda Ministry of Finance and Economic Planning. National Strategy for Transformation (NST1) 2017- 2024 [Internet]. 2018 [cited 2025 Apr 28]. Available from: http://www.minecofin.gov.rw/ Idowu OO. Women-Focused Legislation, Economic Empowerment and Sustainable Development in Africa. In: Akinola AO, Liaga EA, editors. Development and Regional Stability in Africa [Internet]. Cham: Springer Nature Switzerland; 2024 [cited 2025 Apr 28]. p. 107–30. Available from: https://link.springer.com/10.1007/978-3-031-56695-0_7 Jabbi A, Ndow B, Senghore T, Sanyang E, Kargbo JC, Bass P. Prevalence and factors associated with intimate partner violence against women in The Gambia: a population-based analysis. Women Health. 2020 Sep 13;60(8):912–28. Shaikh MA. Prevalence, Correlates, and Trends of Intimate Partner Violence Against Women in Gambia: Results From 2 National Cross-Sectional Health Surveys. Inq J Health Care Organ Provis Financ. 2023 Jan;60:00469580231172879. Meskele M, Khuzwayo N, Taylor M. Intimate partner violence against women living with and without HIV, and the associated factors in Wolaita Zone, Southern Ethiopia: A comparative cross-sectional study. Kabir R, editor. PLOS ONE. 2019 Aug 23;14(8):e0220919. Gubi D, Nansubuga E, Wandera SO. Correlates of intimate partner violence among married women in Uganda: a cross-sectional survey. BMC Public Health. 2020 Dec;20(1):1008. Lencha B, Ameya G, Baresa G, Minda Z, Ganfure G. Intimate partner violence and its associated factors among pregnant women in Bale Zone, Southeast Ethiopia: A cross-sectional study. Brownie SM, editor. PLOS ONE. 2019 May 1;14(5):e0214962. Chollet F. Deep Learning with Python [Internet]. United States of America: y Manning Publications Co; 2018 [cited 2025 May 15]. Available from: https://file.fouladi.ir/courses/deep/books/Fran%C3%A7ois%20Chollet-Deep%20Learning%20with%20Python-Manning%20(2018)-.pdf Xie Y, Zhai Y, Lu G. Evolution of artificial intelligence in healthcare: a 30-year bibliometric study. Front Med [Internet]. 2025;11. Available from: https://dx.doi.org/10.3389/fmed.2024.1505692 Hui V, Constantino RE, Lee YJ. Harnessing Machine Learning in Tackling Domestic Violence—An Integrative Review. Int J Environ Res Public Health. 2023;20(6):4984. Parmigiani G, Barchielli B, Casale S, Mancini T, Ferracuti S. The impact of machine learning in predicting risk of violence: A systematic review. Front Psychiatry [Internet]. 2022;13. Available from: https://dx.doi.org/10.3389/fpsyt.2022.1015914 Lundberg S, Lee SI. A Unified Approach to Interpreting Model Predictions [Internet]. arXiv; 2017 [cited 2025 Apr 29]. Available from: https://arxiv.org/abs/1705.07874 Blagus R, Lusa L. SMOTE for high-dimensional class-imbalanced data. BMC Bioinformatics. 2013;14(1):106. Heseltine-Carp W, Courtman M, Browning D, Kasabe A, Allen M, Streeter A, et al. Machine learning to predict stroke risk from routine hospital data: A systematic review. Int J Med Inf. 2025 Apr;196:105811. Wu Z, Chen H, Ke S, Mo L, Qiu M, Zhu G, et al. Identifying potential biomarkers of idiopathic pulmonary fibrosis through machine learning analysis. Sci Rep [Internet]. 2023;13(1). Available from: https://dx.doi.org/10.1038/s41598-023-43834-z Li G, Li J, Tian Y, Zhao Y, Pang X, Yan A. Machine learning-based classification models for non-covalent Bruton’s tyrosine kinase inhibitors: predictive ability and interpretability. Mol Divers. 2024;28(4):2429–47. Jung Y, Hu J. AK-fold averaging cross-validation procedure. J Nonparametric Stat. 2015;27(2):167–79. Bernert RA, Hilberg AM, Melia R, Kim JP, Shah NH, Abnousi F. Artificial Intelligence and Suicide Prevention: A Systematic Review of Machine Learning Investigations. Int J Environ Res Public Health. 2020 Aug 15;17(16):5929. Reese BM, Chen MS, Nekkanti M, Mulawa MI. Prevalence and Risk Factors of Women’s Past-Year Physical IPV Perpetration and Victimization in Tanzania. J Interpers Violence. 2021 Feb;36(3–4):1141–67. Chikhungu LC, Amos M, Kandala N, Palikadavath S. Married Women’s Experience of Domestic Violence in Malawi: New Evidence From a Cluster and Multinomial Logistic Regression Analysis. J Interpers Violence. 2021 Sep;36(17–18):8693–714. Lasong J, Zhang Y, Muyayalo KP, Njiri OA, Gebremedhin SA, Abaidoo CS, et al. Domestic violence among married women of reproductive age in Zimbabwe: a cross sectional study. BMC Public Health. 2020 Dec;20(1):354. Hossain Md, Asadullah Md, Rahaman A, Miah Md, Hasan M, Paul T, et al. Prediction on Domestic Violence in Bangladesh during the COVID-19 Outbreak Using Machine Learning Methods. Appl Syst Innov. 2021 Oct 13;4(4):77. Rahman R, Khan MdNA, Sara SS, Rahman MdA, Khan ZI. A comparative study of machine learning algorithms for predicting domestic violence vulnerability in Liberian women. BMC Womens Health. 2023 Oct 17;23(1):542. Ikekwuibe IC, Okoror CEM. The pattern and socio-cultural determinants of intimate partner violence in a Nigerian rural community. Afr J Prim Health Care Fam Med [Internet]. 2021 Jun 1 [cited 2025 Apr 28];13(1). Available from: https://phcfm.org/index.php/phcfm/article/view/2435 Additional Declarations No competing interests reported. Supplementary Files Supplementaryfile1.docx Cite Share Download PDF Status: Under Review Version 1 posted Reviews received at journal 15 Nov, 2025 Reviewers agreed at journal 13 Nov, 2025 Reviewers invited by journal 13 Nov, 2025 Editor invited by journal 25 Sep, 2025 Editor assigned by journal 25 Sep, 2025 Submission checks completed at journal 25 Sep, 2025 First submitted to journal 22 Sep, 2025 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-7686865","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":549334762,"identity":"12eb243b-9e34-4874-a7b3-34b48e2ad094","order_by":0,"name":"Musa Nget","email":"data:image/png;base64,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","orcid":"","institution":"Central South University","correspondingAuthor":true,"prefix":"","firstName":"Musa","middleName":"","lastName":"Nget","suffix":""},{"id":549334764,"identity":"63c612cd-a7ec-4af1-9000-917f94998b8b","order_by":1,"name":"Aliane Uwera","email":"","orcid":"","institution":"Central South University","correspondingAuthor":false,"prefix":"","firstName":"Aliane","middleName":"","lastName":"Uwera","suffix":""},{"id":549334766,"identity":"33d1aaa6-5e44-4fe9-8b9e-a0223e72140a","order_by":2,"name":"Eva Moeng","email":"","orcid":"","institution":"Central South University","correspondingAuthor":false,"prefix":"","firstName":"Eva","middleName":"","lastName":"Moeng","suffix":""},{"id":549334768,"identity":"9c3e26cb-68d8-4824-a3c2-7935122add98","order_by":3,"name":"Amadou Barrow","email":"","orcid":"","institution":"University of Florida","correspondingAuthor":false,"prefix":"","firstName":"Amadou","middleName":"","lastName":"Barrow","suffix":""},{"id":549334769,"identity":"ec8fffad-7d01-4e4d-91de-55354a232f24","order_by":4,"name":"Jam Camara","email":"","orcid":"","institution":"Kanazawa University","correspondingAuthor":false,"prefix":"","firstName":"Jam","middleName":"","lastName":"Camara","suffix":""},{"id":549334770,"identity":"b3020856-ac4f-4158-8979-701c498f9b49","order_by":5,"name":"Menen Tsegaw","email":"","orcid":"","institution":"Ambo University","correspondingAuthor":false,"prefix":"","firstName":"Menen","middleName":"","lastName":"Tsegaw","suffix":""}],"badges":[],"createdAt":"2025-09-23 01:38:22","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7686865/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7686865/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":96738418,"identity":"36a126d8-8d9a-468e-87c4-fae0962f6ad6","added_by":"auto","created_at":"2025-11-25 14:43:35","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":368747,"visible":true,"origin":"","legend":"","description":"","filename":"ApplicationofMachineLearninginpredictingintimatepartner.docx","url":"https://assets-eu.researchsquare.com/files/rs-7686865/v1/922fa81c70bcc1be1fe23454.docx"},{"id":96738417,"identity":"0fcba902-02f4-4863-beea-d70f0a5b0f85","added_by":"auto","created_at":"2025-11-25 14:43:35","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":8298,"visible":true,"origin":"","legend":"","description":"","filename":"4367257937ee4bd6ad7b20100356eb66.json","url":"https://assets-eu.researchsquare.com/files/rs-7686865/v1/cd2fe2de24cb915dd72f2e06.json"},{"id":96914364,"identity":"5ac2eebf-efd3-4707-9ce8-d3bce28f1b30","added_by":"auto","created_at":"2025-11-27 14:05:49","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":483689,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementaryfile1.docx","url":"https://assets-eu.researchsquare.com/files/rs-7686865/v1/952103c6c448bb3726cd206c.docx"},{"id":96914133,"identity":"ac5a778a-c3bd-41dd-b78c-170e221478c5","added_by":"auto","created_at":"2025-11-27 14:05:29","extension":"xml","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":104796,"visible":true,"origin":"","legend":"","description":"","filename":"4367257937ee4bd6ad7b20100356eb661enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7686865/v1/33d829d6788772e5ad09edad.xml"},{"id":96914011,"identity":"b27f7c6e-92db-4420-a029-c07dc4e0be56","added_by":"auto","created_at":"2025-11-27 14:05:07","extension":"png","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":99022,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7686865/v1/c8f652c4ec2fc5eec94fcf93.png"},{"id":96738424,"identity":"1bcd7ccd-b649-4e4a-872e-357b120ffe38","added_by":"auto","created_at":"2025-11-25 14:43:35","extension":"jpeg","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":611400,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7686865/v1/1102f17af672deebaa64b267.jpeg"},{"id":96914010,"identity":"e2c2e6ce-03b5-4eb6-8160-43d1a5548c82","added_by":"auto","created_at":"2025-11-27 14:05:07","extension":"jpeg","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":334804,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7686865/v1/70a03f7833645bc650032e31.jpeg"},{"id":96738427,"identity":"9c6c316b-794f-46ce-bd9d-9a8d1bcb52f0","added_by":"auto","created_at":"2025-11-25 14:43:35","extension":"png","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":23179,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7686865/v1/4e4725505823ce5f07224414.png"},{"id":96738430,"identity":"7b3c0efa-ad2a-43ce-8b47-8d97d4f2c159","added_by":"auto","created_at":"2025-11-25 14:43:35","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":120311,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7686865/v1/1161d7e3d754153acc0c6e6c.png"},{"id":96914229,"identity":"f8de57d0-f3dc-48bd-89bd-72127df76bd2","added_by":"auto","created_at":"2025-11-27 14:05:37","extension":"png","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":53647,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7686865/v1/33162bda31b0bec52c9b6d10.png"},{"id":96738432,"identity":"a2010da9-d737-4a17-8415-05541bf12650","added_by":"auto","created_at":"2025-11-25 14:43:35","extension":"xml","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":103431,"visible":true,"origin":"","legend":"","description":"","filename":"4367257937ee4bd6ad7b20100356eb661structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7686865/v1/681399a2ae4a4f7da0bd96d7.xml"},{"id":96738431,"identity":"b107a47d-4f74-4f5d-bd50-949aadbfbc93","added_by":"auto","created_at":"2025-11-25 14:43:35","extension":"html","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":111105,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7686865/v1/64da753247becae657065d1b.html"},{"id":96738416,"identity":"1597826b-7758-4566-b8a3-fbce1938ddc9","added_by":"auto","created_at":"2025-11-25 14:43:35","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":146803,"visible":true,"origin":"","legend":"\u003cp\u003eWorkflow of variable selection and model development. RF: Random forest, LR: Logistic regression, Lasso: least absolute shrinkage and selection operator, SVM: Support vector machine, CART: Classification and Regression Trees\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7686865/v1/9ec42b6f7571ba2391cd50bb.png"},{"id":96738422,"identity":"c6263710-d01b-4937-97ff-271c65a2e4a0","added_by":"auto","created_at":"2025-11-25 14:43:35","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":551184,"visible":true,"origin":"","legend":"\u003cp\u003eFig 1: Feature importance for the top 15 features across all ML models\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7686865/v1/7502edbc0070fe79295d0316.png"},{"id":96738420,"identity":"6e997c79-e61d-4ac3-8542-d831c637862c","added_by":"auto","created_at":"2025-11-25 14:43:35","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":335349,"visible":true,"origin":"","legend":"\u003cp\u003eFig 3: ROC Curves for IPV Prediction Models\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7686865/v1/fb49d9a211f110ca044e577b.png"},{"id":97248589,"identity":"ec031752-3dc7-4018-adcb-7c7e2ea36ea4","added_by":"auto","created_at":"2025-12-02 13:03:48","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1864667,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7686865/v1/446950a1-9ba8-4b8f-9094-ce9dcf133a11.pdf"},{"id":96738419,"identity":"60e6f944-90c1-438e-9a0b-984511e4dd3e","added_by":"auto","created_at":"2025-11-25 14:43:35","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":483689,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementaryfile1.docx","url":"https://assets-eu.researchsquare.com/files/rs-7686865/v1/fab855e7d87938d01489df0e.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Application of Machine Learning in predicting intimate partner-based violence among reproductive age women in Rwanda","fulltext":[{"header":"Introduction","content":"\u003cp\u003eIntimate partner violence (IPV) is a serious global public health issue that predominantly affects women, although men can also be victims\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. According to the World Health Organization (WHO), IPV is defined as acts by a current or former intimate partner that cause physical, sexual, or psychological harm, including physical aggression, sexual coercion, psychological abuse, and controlling behaviours\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. The Center for Disease Control and Prevention (CDC) typically categorizes IPV into four types: sexual violence, physical violence, threats, and psychological or emotional abuse\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eGlobally, the prevalence of IPV remains high, especially in low-income countries. Approximately 27% of women aged 15\u0026ndash;49 in intimate relationships report experiencing IPV, and IPV accounts for 38% of female homicides worldwide\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. Despite its widespread nature, IPV data and prevention strategies remain limited, particularly in low- and middle-income countries (LMICs), with sub-Saharan Africa bearing a significant burden\u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e. Previous research has shown that women who have experienced physical or sexual intimate partner violence are at risk of developing physical problems\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e,\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e, as well as mental health issues such as depression, anxiety, posttraumatic stress disorder, suicide, and being drunkards\u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eThe Global Burden of Disease highlights that the highest prevalence of IPV occurs in central sub-Saharan Africa (32%) and Oceania (29%), followed by eastern sub-Saharan Africa (24%), and South Asia (19%)\u003csup\u003e7\u003c/sup\u003e. In Rwanda, the prevalence of IPV was reported to be 40.21%\u003csup\u003e8\u003c/sup\u003e, with a higher percentage of women (49.6%) seeking help\u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e. The causes and effects of IPV on women, especially in Africa, are subject to various speculations and cultural norms.\u003c/p\u003e\u003cp\u003eRwanda is often regarded as a leader in Africa in promoting gender equality, mainly due to its progressive policies and strong legal frameworks that target gender-based violence, including IPV\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e. Like other countries around the world, Rwanda has rules and policies to protect women from violence, unlike other countries in the region. For example, in 2008, the Rwandan government passed the Prevention and Punishment of Gender-based Violence Law, which encompasses all types of violence and has a minimum prison penalty of six months. These initiatives have contributed to significant reductions in IPV and improvements in women's rights\u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eThe previous studies conducted across different regions highlighted age, education level, alcohol and substance use, childhood exposure to violence, the duration of relationships, and partner\u0026rsquo;s controlling behaviour as the common emerging factors associated with IPV. These factors vary widely across cultural and socioeconomic settings. A recent study revealed that intimate partners with a certain level of education and those who had completed primary or higher education were more likely to report IPV\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e,\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e. Additionally, a cross-sectional study reported that women whose partners consumed alcohol reported IPV 2.3 times higher than their counterparts\u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e. Gubi et al found that partners with longer periods of relationship had higher odds of experiencing IPV\u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e. Moreover, growing up in a family that normalized violence was associated with IPV, whereby people who were exposed to violence during their childhood are prone to acceptance of IPV and are immersed in the mire of recurrent IPV\u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e. Furthermore, various studies found that women in relationships with men who had controlling behaviours or had been involved in any form of aggressive behaviour were highly likely to experience IPV \u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e,\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eAlthough efforts to reduce intimate partner violence (IPV) have been put in place by different organizations and countries, it continues to pose a significant public health concern that necessitates more targeted interventions\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e. Despite IPV-related injuries continuing to be a leading cause of injury and death worldwide, data to inform prevention strategies remain limited in low and middle-income countries (LMICs), especially in sub-Saharan Africa\u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eMachine learning is a subfield of artificial intelligence that focuses on creating systems that learn patterns from data rather than being explicitly programmed with rules. It allows models to automatically learn from historical data to make accurate forecasts or classifications about future or unseen events\u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e. Over the past decade, machine learning (ML) has emerged as a robust and promising research methodology in healthcare research throughout the early detection, diagnosis, treatment, and prognosis evaluation\u003csup\u003e\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e. Researchers have applied ML models to study the risk factors for aggressive episodes of various types of violence. However, the development of ML algorithms is still in its infancy\u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e. Thus, more research and investments are required, preferably in large and prospective groups, to boost the application of ML models in health care and clinical practice\u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e. Therefore, this study seeks to develop and validate a robust machine learning model to predict IPV and its predictors in Rwanda, thus enabling timely interventions to prevent intimate partner violence.\u003c/p\u003e"},{"header":"Method","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003eStudy design and data source\u003c/h2\u003e\u003cp\u003eThe study used data from the Rwanda Demographic and Health Survey 2019\u0026ndash;2020, conducted by the National Institute of Statistics of Rwanda (NISR) in partnership with the Global Demographic and Health Surveys Program. A two-stage stratified cluster sample design was used to collect data from 14,634 women within the reproductive age group of 15\u0026ndash;49 years in Rwanda. Initially, clusters were selected from enumeration areas, followed by the selection of households within these clusters. RDHS 2019\u0026ndash;2020 was a cross-sectional, nationally representative survey comprising multiple datasets, including but not limited to the IPV dataset. This study used the individual file record (RWFW81), which includes data on intimate partner violence.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eOutcome variable\u003c/h3\u003e\n\u003cp\u003eEver-witnessed IPV was the dependent variable in the present study. Two categories, \u0026ldquo;Yes\u0026rdquo; (experienced IPV) and \u0026ldquo;No\u0026rdquo; (did not experience IPV), were used to classify the dependent variable. One indicates \u0026ldquo;Yes\u0026rdquo;, whereas zero indicates \u0026ldquo;No\u0026rdquo;. Physical violence (pushing, slapping, punching with a fist or something harmful, kicking or beating, threats, or strangulation or burning). Emotional violence (humiliation in front of others, insults and negative self-perception, intimidation, isolation from friends and family, jealousy and accusations of infidelity, and control over movements and activities). Sexual violence (forced sexual acts against her will, physical injuries resulting from sexual violence, forced sexual intercourse during pregnancy, experiences involving multiple perpetrators, or forced sexual acts perceived as humiliation).\u003c/p\u003e\n\u003ch3\u003eFeatures\u003c/h3\u003e\n\u003cp\u003eThis study explored multiple factors that influence intimate partner violence through three key categories. Sociodemographic features included residence type, age, education, marital status, region, religion, wealth, travel pattern, and cohabitation status. Behavioural features examined included relationship dynamics, sexual autonomy, alcohol use, communication monitoring, and reproductive choices, including forced sexual initiation and abortion history. The psychosocial features focused on decision-making autonomy regarding finances, healthcare, social interactions, and childhood exposure to violence. Healthcare and structural vulnerability factors specifically address physical harm during pregnancy caused by the former or previous partner.\u003c/p\u003e\n\u003ch3\u003eData pre-processing\u003c/h3\u003e\n\u003cp\u003eThis study involved various steps to prepare the RWDHS 2019\u0026ndash;2020 dataset. The dataset consisted of 60 features with a single target feature. The data underwent a cleaning process, and missing values were handled using the K-nearest neighbor approach and multiple imputation. Initially, to prepare the data for the model, it is necessary to convert the input and output features into numerical values. The researchers used a one-hot encoder to encode the categorical variables found in the dataset. This technique transforms each categorical value into a new column with one-hot encoding, and the label values are created as new columns with values of either 1 or 0. Secondly, we used multiple imputation techniques to address missing values and incomplete raw data. As detailed in the supplementary figures, a breakdown of the percentage of missing values for each feature is as follows: argues with partner 4582 (31.3%), jealous partner 291(2%), burns food 2632(18%), insists on knowing respondent\u0026rsquo;s whereabouts 737(5%), refuses to have sex 4550 (31.1%), partner accusation of unfaithfulness 291(2%), rural residence 11,083(75.7%), alcohol consumption 1230(8.4%), residing 6511 (44.5%), wife looking at the phone (missing), child embarrasses family (missing), and sex outside marriage (missing). Subsequently, duplicates were screened and eliminated from the final dataset.\u003c/p\u003e\n\u003ch3\u003eFeature selection\u003c/h3\u003e\n\u003cp\u003eSHAP (Shapley Additive Explanations) is a technique for elucidating machine learning model predictions by calculating feature contributions derived from Shapley values in cooperative game theory\u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e. It assigns the disparity between a model's forecast and baseline value to each feature, guaranteeing coherence and local precision in the explanation.\u003c/p\u003e\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003eImbalanced data handling\u003c/h2\u003e\u003cp\u003eImbalanced data in machine learning poses a significant problem because it can result in bias that favors the majority class. This difficulty typically arises when the distribution of classes within a dataset is unbalanced. To achieve dataset balance in the current study, we used the Synthetic Minority Over-sampling Technique (SMOTE). SMOTE is an oversampling technique that generates synthetic instances of the minority class through the computation of existing minority samples. This method balances class distribution in the training dataset, enhancing the classifier's capacity to learn from imbalanced data\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e. We evaluated its effectiveness by comparing the chosen supervised machine learning algorithms, emphasizing the accuracy and AUC. This study illustrates that the use of balanced sampling techniques is essential.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eModel development\u003c/h3\u003e\n\u003cp\u003eThis study used machine learning algorithms, including logistic regression (LR), Classification and Regression Trees (CART), Least Absolute Shrinkage and Selection Operator (LASSO), Support Vector Machine (SVM), and Random Forest (RF) to predict intimate partner violence.\u003c/p\u003e\u003cp\u003eLogistic regression (LR) is one of the earliest and most commonly used classification algorithms in machine learning that predicts the probability of a binary outcome based on one or more input variables by modeling the log-odds of the outcome as a linear combination of the input features \u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e,\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e. LR was used as a foundational model for selecting input features in the prediction model, then feature selection was performed using the Shapley additive explanation (SHAP) method to identify the most relevant predictors. Subsequently, we used four ML algorithms (CART, LASSO, SVM, RF) to evaluate their effectiveness in predicting violent behavior among intimate partners.\u003c/p\u003e\u003cp\u003eClassification and Regression Trees is a decision tree algorithm for classification and regression tasks. It splits the dataset into subsets based on feature values that result in the most homogeneous subgroups\u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eLeast Absolute Shrinkage and Selection Operator is a linear regression algorithm with regularization applied to high-dimensional analysis\u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eSupport Vector Machine (SVM) is a supervised classification algorithm that aims to find the optimal decision boundary that maximizes the margin between classes in the data points by transforming input data into a high-dimensional feature space using kernel functions. This method allows the model to handle non-linear relationships\u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e. SVM has a strong theoretical underpinning and has revealed an effective performance on various tasks\u003csup\u003e\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e .\u003c/p\u003e\u003cp\u003eRandom forest (RF) is a supervised ensemble learning method that generates a single output or prediction by integrating the outcomes of various decision trees. It reduces overfitting through random sampling of both data points and features\u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\n\u003ch3\u003eModel evaluation (Prediction Model Evaluation)\u003c/h3\u003e\n\u003cp\u003eThe researchers used k-fold cross-validation (k\u0026thinsp;=\u0026thinsp;10) to guarantee the strength and applicability of the machine learning model for predicting intimate partner violence (IPV) and to assess its performance. This approach reduces the risk of overfitting by dividing the dataset into k subsets, where the model is systematically trained on k-1 folds and validated on the left-out fold \u003csup\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e. We divided the training set into ten equal subsets, one for validation and the remaining nine for training. Upon completion of cross-validation, the accuracy and kappa metrics were utilized to produce the final models for each algorithm. The evaluation of these models on the test set involved various performance metrics, such as F1 score, AUC, accuracy, precision, sensitivity, and specificity. The model with the highest AUC value was identified as the optimal choice.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eFigure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e: Workflow of variable selection and model development. RF: Random forest, LR: Logistic regression, Lasso: least absolute shrinkage and selection operator, SVM: Support vector machine, CART: Classification and Regression Trees\u003c/p\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003eParticipant\u0026rsquo;s characteristics\u003c/h2\u003e\u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e presents the characteristics of the study population after excluding participants who did not pertain to intimate partner violence. The study population comprised 14,634 individuals, with the majority residing in rural areas (75.7%) and a mean age of 29.1 (\u0026plusmn;\u0026thinsp;9.9) years. Nearly half (46.2%) were Protestants, followed by Catholics (37.6%). Most participants had partners with a primary education (58.1%), while 9.2% had no education. Marital status varied, with 41.4% never in union, 32.2% married, and 17.7% living with a partner. Geographically, participants were distributed across all regions, with the highest proportion in the East (24.8%). The wealth distribution was relatively even, with the poorest (19.4%) and richer (19.7%) categories representing almost similar proportions.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eSociodemographic characteristics of the study population\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"4\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFeatures\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eCategory\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eFrequency (n)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003ePercentage (%)\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eResidence\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eRural\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e11,083\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e75.7\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eUrban\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e3,551\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e24.3\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAge*\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e15\u0026ndash;19\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e3,308\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e22.6\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e20\u0026ndash;24\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e2,424\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e16.6\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e25\u0026ndash;29\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e2,047\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e14.0\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e30\u0026ndash;34\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e2,095\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e14.3\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e35\u0026ndash;39\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e2,043\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e14.0\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e40\u0026ndash;44\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1,487\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e10.2\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e45\u0026ndash;49\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1,230\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e8.4\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePartner\u0026rsquo;s education\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eHigher\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e672\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e4.6\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNo education\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1,352\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e9.2\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003ePrimary\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e8,500\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e58.1\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSecondary\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e4,110\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e28.1\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMarital status\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eDivorced\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e379\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e2.6\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eLiving with partner\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e2,584\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e17.7\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMarried\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e4,706\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e32.2\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNever in union\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e6,060\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e41.4\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSeparated\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e527\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e3.6\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eWidowed\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e378\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e2.6\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRegion\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eEast\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e3,625\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e24.8\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eKigali\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1,921\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e13.1\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNorth\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e2,294\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e15.7\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSouth\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e3,482\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e23.8\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eWest\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e3,312\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e22.6\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eReligion\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eAdventist\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1,842\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e12.6\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eCatholic\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e5,506\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e37.6\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMuslim\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e287\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e2.0\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNo religion\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e114\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.8\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eProtestant\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e6,754\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e46.2\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eTraditional\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.0\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eUnknown\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e130\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.9\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eWealth index\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMiddle\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e2,709\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e18.5\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003ePoorer\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e2,707\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e18.5\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003ePoorest\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e2,844\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e19.4\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eRicher\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e2,884\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e19.7\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eRichest\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e3,490\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e23.8\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTrips in the last 12 months\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e90+\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e21\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.1\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNone\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e8,440\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e57.7\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eUnknown\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e6,173\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e42.2\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e\u003cp\u003eCurrently residing with partner\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e6,511\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e44.5\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e8123\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e55.5\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e\u003cp\u003eIntimate Partner Violence\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eExperience IPV\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e149\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1.02\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNo IPV\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e38\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.26\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eNA\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e14447\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e98.72\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003ctfoot\u003e\u003ctr\u003e\u003ctd colspan=\"4\"\u003e* Mean (standard deviation): 29.13\u0026thinsp;\u0026plusmn;\u0026thinsp;9.92\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\u003ch2\u003eFeature selection\u003c/h2\u003e\u003cp\u003eThe analysis of feature importance across five machine learning models: Logistic Regression, Lasso, Random Forest, CART, and SVM, revealed key predictors associated with IPV (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e, supplementary file Fig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e-S5). The most prominent predictor, which showed the highest importance score across almost all models, was arguing with a partner and was particularly influential in the logistic regression model (9.995), indicating that frequent disputes strongly correlate with the outcome. Other significant predictors in the logistic regression model included a jealous partner (2.794) and burning food (1.629), with jealousy showing notable importance in the Random Forest (0.044) and CART (0.109) models. Controlling behaviors such as insisting on knowing the respondent\u0026rsquo;s whereabouts (1.024) and accusations of unfaithfulness (0.627) were consistently influential across multiple models. Sexual behavior issues, such as refusing to have sex (0.645) and engaging in sex outside marriage (0.247), also contributed, albeit to a lesser extent. For other factors, residence (0.534) and alcohol consumption (0.403) had relatively low but non-negligible importance, with alcohol use appearing in several models. Methodologically, Logistic Regression and Random Forest models assigned greater weights to these features compared to the other models, suggesting differences in how these models prioritize predictors.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e\u003ch2\u003eDevelopment and evaluation of the model\u003c/h2\u003e\u003cp\u003eThe ML models demonstrated varying performance in predicting IPV, with tree-based methods (Random Forest and CART) outperforming other methods (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e and Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e3\u003c/span\u003e). Random Forest achieved perfect discrimination (AUC\u0026thinsp;=\u0026thinsp;1.000) and near-perfect accuracy (0.991), precision (0.995), recall (0.986), and specificity (0.995), indicating exceptional ability to identify both true positives and negatives. Similarly, CART exhibited strong performance (AUC\u0026thinsp;=\u0026thinsp;0.986, accuracy\u0026thinsp;=\u0026thinsp;0.986) with balanced precision (0.986) and recall (0.985). Logistic regression also performed well (AUC\u0026thinsp;=\u0026thinsp;0.981, accuracy\u0026thinsp;=\u0026thinsp;0.962), but had a slightly lower recall (0.928), suggesting minor under-detection of true IPV cases. In contrast, Lasso showed moderate discrimination (AUC\u0026thinsp;=\u0026thinsp;0.821) but poor specificity (0.24), reflecting a high false-positive rate despite its high recall (0.97), whereas the SVM model performed inadequately (AUC\u0026thinsp;=\u0026thinsp;0.569, recall\u0026thinsp;=\u0026thinsp;0.293), failing to detect true IPV cases. These highlight Random Forest and CART as the most robust models for IPV prediction by combining high accuracy with balanced sensitivity and specificity. Logistic regression remains a viable alternative, while LASSO and SVM are unsuitable because of significant trade-offs in performance.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eEvaluation of the model\u0026rsquo;s performance\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"6\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eModels\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eAccuracy\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003ePrecision\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eRecall\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eSpecificity\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003eF1 score\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLogistic Regression\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.962\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.995\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.928\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.995\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.960\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRandom Forest\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.991\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.995\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.986\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.995\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.991\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLasso\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.605\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.561\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.970\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.240\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.710\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSVM\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.610\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.800\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.293\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.927\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.429\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCART\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.986\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.986\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.985\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.986\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.986\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eMachine learning (ML) has emerged as a robust analysis method because of its ability to analyze large datasets and model complex relationships between risk factors and outcomes\u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e. The current study represents a significant contribution to the application of machine learning for predicting IPV in Rwanda. Using socio-demographic, sociocultural, psychosocial, behavioural, sexual, healthcare, and structural features retrieved from the 2019\u0026ndash;2020 Rwanda Demographic Health Survey, this study was able to predict the IPV status of individuals. After comparing five different ML algorithms, Random Forest emerged as the most effective algorithm for predicting IPV status. The research utilized SMOTE to overcome the dataset imbalance problem, thereby significantly improving the performance and reliability of the prediction model. To increase the interpretability of our predictions, we also performed a SHAP analysis to determine and measure each feature's contribution to our model's results. The model identified 15 significant factors among all variables considered in this study: arguing with partner, jealous partner, burning food, insistence on knowing the respondent's whereabouts, refusal to have sex, partner's accusations of unfaithfulness, place of residence, hurt during pregnancy by boyfriend, humiliation by previous partner, being physically forced to have sex by a previous partner, alcohol consumption, living with partner, wife looking through husband's phone, child embarrassing the family, and sex outside marriage.\u003c/p\u003e\u003cp\u003eThe results revealed that IPV was 1.02% among Rwandans. The findings contradict the global results, where a study conducted by the World Health Organization among 161 countries around the world found that 30% of women experience IPV in their lifetime\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. Additionally, the current findings were lower than those of studies conducted in East African regions, 56% of Uganda\u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e, 44% Tanzania\u003csup\u003e27,\u003c/sup\u003e and also lower than those of studies conducted in other African regions (40% Gambia\u003csup\u003e11\u003c/sup\u003e, 42% Malawi\u003csup\u003e28\u003c/sup\u003e, 42.7% Zimbabwe\u003csup\u003e29\u003c/sup\u003e. The higher figures of IPV in African countries could be attributed to social and cultural acceptance of IPV as part of social interactions in relationships. Additionally, the difference in results could be attributed to the difference in research methodologies used among these studies; however, this implies that IPV remains a public issue, especially in African countries. Although some countries have implemented violence prevention policies, more effective policies and awareness are needed to reduce IPV.\u003c/p\u003e\u003cp\u003eIn the current study, most algorithms demonstrated high performance, particularly Random Forest, CART, and logistic regression, which showed strong and consistent results across all evaluation metrics. Random Forest achieved the highest accuracy and sensitivity scores of 99.1% and 98.6%, respectively. A previous study on violence among women similarly identified random forest as the best predictive model, with an accuracy and sensitivity score of 77% and 78%, respectively\u003csup\u003e\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e. On the other hand, a study that used the Liberian 2019 DHS dataset and seven different machine learning algorithms found that decision tree and CatBoost outperformed the others in identifying women at risk of domestic violence, with an accuracy rate of 82%, while random forest revealed an accuracy rate of 81%\u003csup\u003e31\u003c/sup\u003e. This can be attributed to its ensemble learning approach, which lowers overfitting while increasing generalizability and handling imbalanced secondary data. Therefore, our findings emphasize the predictive power of Random for IPV.\u003c/p\u003e\u003cp\u003eHossain et al. revealed that machine learning algorithms, particularly random forests, successfully predicted domestic violence. The random forest method provided better results across all metrics\u003csup\u003e\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e, similar to the results of the current study. With an AUC of 94%, Random Forest outperformed other algorithms in their research\u003csup\u003e\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e. The random forest model was the best IPV predictive model in this study, with an AUC of 100%. The higher AUC of our model emphasizes that random forest accuracy predicts intimate partner violence by successfully distinguishing between positive and negative cases.\u003c/p\u003e\u003cp\u003eOur study found several significant predictors of IPV; these features represent both interpersonal conflict and cultural standards that influence gender roles in justifying IPV. For instance, arguing with a partner, being jealous of a partner, burning food, refusal to have sex, and sex outside marriage justify the likelihood of experiencing IPV. Notably, a previous study emphasized that victims justify IPV under specific circumstances (refusal to have sex and partner infidelity) based on traditional acceptance\u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e. This suggests that more women are at risk of experiencing IPV, as justification for such violence is frequently based on societal norms, tolerance, and those who may try to seek help are stigmatized and viewed as failing to be submissive in marriage.\u003c/p\u003e\u003cdiv id=\"Sec16\" class=\"Section2\"\u003e\u003ch2\u003eStrengths and limitations\u003c/h2\u003e\u003cp\u003eThe strengths of this study lie in its use of a national dataset and advanced ML techniques to identify key predictors of IPV. The current research uses SMOTE, which helps balance the class distribution and improve the model's ability to generalize and predict minority classes accurately. Moreover, we employed SHAP to explain why RF exhibited the best performance among ML algorithms. SHAP solves the problem of multicollinearity by considering not only the influence of a single feature but also the synergy between features. However, significant improvements have been made in both the features and results, resulting in a more reliable model. This study has some limitations, including reliance on self-reported data, which may introduce bias, and the cross-sectional nature of the survey, which precludes causal inferences.\u003c/p\u003e\u003c/div\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThis study applied ML models to predict IPV among reproductive-aged women in Rwanda, using data from the 2019\u0026ndash;2020 Rwanda Demographic and Health Survey. The Random Forest algorithm outperformed other models (logistic regression, CART, Lasso, SVM) with exceptional accuracy (99.1%) and sensitivity (98.6%), effectively identifying key predictors such as interpersonal conflicts (e.g., arguing with a partner, jealousy), controlling behaviours, and cultural justifications for violence (e.g., refusal to have sex, accusations of infidelity). Machine learning models that incorporate various features may serve as an effective approach to address IPV and could help in early intervention and prevention by strengthening policy enforcement, integrating predictive tools into healthcare systems, and launching community awareness campaigns to address the underlying cultural norms.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eADASYN: Adaptive Synthetic Sampling\u003c/p\u003e\n\u003cp\u003eAUC: Area Under the ROC Curve\u003c/p\u003e\n\u003cp\u003eCART: Classification and Regression Trees\u003c/p\u003e\n\u003cp\u003eCDC: The Center for Disease Control and Prevention\u003c/p\u003e\n\u003cp\u003eIPV: Intimate partner violence\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eLASSO: Least Absolute Shrinkage and Selection Operator\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eLMICs: low- and middle-income countries\u003c/p\u003e\n\u003cp\u003eLR: logistic regression\u003c/p\u003e\n\u003cp\u003eML: Machine learning \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eNISR: National Institute of Statistics of Rwanda \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eRF: Random forest\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eROC: Receiver Operating Characteristic\u003c/p\u003e\n\u003cp\u003eRWDHS - Rwanda Demographic and Health Survey\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSHAP: Shapley Additive Explanations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSMOTE: Synthetic Minority Oversampling Technique\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eSVM: Support Vector Machine\u003c/p\u003e\n\u003cp\u003eWHO: World Health Organization\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eData availability statement\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe data that support the findings of this study are openly available in The DHS Program at https://dhsprogram.com/data/available-datasets.cfm, reference country Rwanda.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics statement\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eEthics approval for this study was not required as\u0026nbsp;we used secondary data that is publicly accessible. The survey, in its original format, obtained ethical approval from the Rwanda Research Ethics Committee (RWIR81FL.SAV). Source: Rwanda 2019/2020 Demographic Health Survey, October 2024 (https://www.dhsprogram.com/data/available-datasets.cfm). We secured authorization from the DHS Program to utilize the Rwanda DHS dataset in our research. The DHS program adheres to the protocol of informed consent and maintains the principles of anonymity and confidentiality under the regulations governing data utilization. Additional information concerning DHS data and ethical standards can be found at: http://goo.gl/ny8T6X.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eMN conceptualised the study, MN, UA, EM and AB analysis of data, MN, UA and EM wrote the introduction, results and discussion, JC, AB and MT and critically reviewed the manuscript for its intellectual content. MN and UA had final responsibility to submit for publication. MN and UA accept full responsibility for the finished work and/or the conduct of the study, had access to the data and controlled the decision to publish.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that no financial assistance was obtained for the research, authoring, and/or publication of this paper.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003cstrong\u003eAcknowledgements\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors express their gratitude to the MEASURE DHS project for providing free access to source data and for their help.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflict of interest\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors affirm that there were no financial or business ties that could be interpreted as potential conflicts of interest during the research.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eWorld Health Organization. Violence Against Women Prevalence Estimates 2018: Global, Regional and National Prevalence Estimates for Intimate Partner Violence Against Women and Global and Regional Prevalence Estimates for Non-Partner Sexual Violence Against Women. 1st ed. Geneva: World Health Organization; 2021. 1 p.\u003c/li\u003e\n\u003cli\u003eIdoko P, Ogbe E, Jallow O, Ocheke A. Burden of intimate partner violence in The Gambia - a cross sectional study of pregnant women. Reprod Health. 2015 Dec;12(1):34.\u003c/li\u003e\n\u003cli\u003eBlair KJ, De Virgilio M, Dissak-Delon FN, Dang LE, Christie SA, Carvalho M, et al. Associations between social determinants of health and interpersonal violence-related injury in Cameroon: a cross-sectional study. BMJ Glob Health. 2022 Jan;7(1):e007220.\u003c/li\u003e\n\u003cli\u003eCheung M, Leung P, Tsui V. Asian Male Domestic Violence Victims: Services Exclusive for Men. J Fam Violence. 2009 Oct;24(7):447\u0026ndash;62.\u003c/li\u003e\n\u003cli\u003eKapiga S, Harvey S, Muhammad AK, St\u0026ouml;ckl H, Mshana G, Hashim R, et al. Prevalence of intimate partner violence and abuse and associated factors among women enrolled into a cluster randomised trial in northwestern Tanzania. BMC Public Health. 2017 Dec;17(1):190.\u003c/li\u003e\n\u003cli\u003eCurry SJ, Krist AH, Owens DK, Barry MJ, Caughey AB, Davidson KW, et al. Screening for Intimate Partner Violence, Elder Abuse, and Abuse of Vulnerable Adults: US Preventive Services Task Force Final Recommendation Statement. JAMA. 2018 Oct 23;320(16):1678.\u003c/li\u003e\n\u003cli\u003eAyebeng C, Dickson KS, Ameyaw EK, Adde KS, Paintsil JA, Yaya S. Influence of type of violence on women\u0026rsquo;s help-seeking behaviour: Evidence from 10 countries in sub-Saharan Africa. Pretorius D, editor. PLOS ONE. 2024 Mar 8;19(3):e0297308.\u003c/li\u003e\n\u003cli\u003eTessema ZT, Gebrie WM, Tesema GA, Alemneh TS, Teshale AB, Yeshaw Y, et al. Intimate partner violence and its associated factors among reproductive-age women in East Africa:-A generalized mixed effect robust poisson regression model. Paul P, editor. PLOS ONE. 2023 Aug 18;18(8):e0288917.\u003c/li\u003e\n\u003cli\u003eRwanda Ministry of Finance and Economic Planning. National Strategy for Transformation (NST1) 2017- 2024 [Internet]. 2018 [cited 2025 Apr 28]. Available from: http://www.minecofin.gov.rw/\u003c/li\u003e\n\u003cli\u003eIdowu OO. Women-Focused Legislation, Economic Empowerment and Sustainable Development in Africa. In: Akinola AO, Liaga EA, editors. Development and Regional Stability in Africa [Internet]. Cham: Springer Nature Switzerland; 2024 [cited 2025 Apr 28]. p. 107\u0026ndash;30. Available from: https://link.springer.com/10.1007/978-3-031-56695-0_7\u003c/li\u003e\n\u003cli\u003eJabbi A, Ndow B, Senghore T, Sanyang E, Kargbo JC, Bass P. Prevalence and factors associated with intimate partner violence against women in The Gambia: a population-based analysis. Women Health. 2020 Sep 13;60(8):912\u0026ndash;28.\u003c/li\u003e\n\u003cli\u003eShaikh MA. Prevalence, Correlates, and Trends of Intimate Partner Violence Against Women in Gambia: Results From 2 National Cross-Sectional Health Surveys. Inq J Health Care Organ Provis Financ. 2023 Jan;60:00469580231172879.\u003c/li\u003e\n\u003cli\u003eMeskele M, Khuzwayo N, Taylor M. Intimate partner violence against women living with and without HIV, and the associated factors in Wolaita Zone, Southern Ethiopia: A comparative cross-sectional study. Kabir R, editor. PLOS ONE. 2019 Aug 23;14(8):e0220919.\u003c/li\u003e\n\u003cli\u003eGubi D, Nansubuga E, Wandera SO. Correlates of intimate partner violence among married women in Uganda: a cross-sectional survey. BMC Public Health. 2020 Dec;20(1):1008.\u003c/li\u003e\n\u003cli\u003eLencha B, Ameya G, Baresa G, Minda Z, Ganfure G. Intimate partner violence and its associated factors among pregnant women in Bale Zone, Southeast Ethiopia: A cross-sectional study. Brownie SM, editor. PLOS ONE. 2019 May 1;14(5):e0214962.\u003c/li\u003e\n\u003cli\u003eChollet F. Deep Learning with Python [Internet]. United States of America: y Manning Publications Co; 2018 [cited 2025 May 15]. Available from: https://file.fouladi.ir/courses/deep/books/Fran%C3%A7ois%20Chollet-Deep%20Learning%20with%20Python-Manning%20(2018)-.pdf\u003c/li\u003e\n\u003cli\u003eXie Y, Zhai Y, Lu G. Evolution of artificial intelligence in healthcare: a 30-year bibliometric study. Front Med [Internet]. 2025;11. Available from: https://dx.doi.org/10.3389/fmed.2024.1505692\u003c/li\u003e\n\u003cli\u003eHui V, Constantino RE, Lee YJ. Harnessing Machine Learning in Tackling Domestic Violence\u0026mdash;An Integrative Review. Int J Environ Res Public Health. 2023;20(6):4984.\u003c/li\u003e\n\u003cli\u003eParmigiani G, Barchielli B, Casale S, Mancini T, Ferracuti S. The impact of machine learning in predicting risk of violence: A systematic review. Front Psychiatry [Internet]. 2022;13. Available from: https://dx.doi.org/10.3389/fpsyt.2022.1015914\u003c/li\u003e\n\u003cli\u003eLundberg S, Lee SI. A Unified Approach to Interpreting Model Predictions [Internet]. arXiv; 2017 [cited 2025 Apr 29]. Available from: https://arxiv.org/abs/1705.07874\u003c/li\u003e\n\u003cli\u003eBlagus R, Lusa L. SMOTE for high-dimensional class-imbalanced data. BMC Bioinformatics. 2013;14(1):106.\u003c/li\u003e\n\u003cli\u003eHeseltine-Carp W, Courtman M, Browning D, Kasabe A, Allen M, Streeter A, et al. Machine learning to predict stroke risk from routine hospital data: A systematic review. Int J Med Inf. 2025 Apr;196:105811.\u003c/li\u003e\n\u003cli\u003eWu Z, Chen H, Ke S, Mo L, Qiu M, Zhu G, et al. Identifying potential biomarkers of idiopathic pulmonary fibrosis through machine learning analysis. Sci Rep [Internet]. 2023;13(1). Available from: https://dx.doi.org/10.1038/s41598-023-43834-z\u003c/li\u003e\n\u003cli\u003eLi G, Li J, Tian Y, Zhao Y, Pang X, Yan A. Machine learning-based classification models for non-covalent Bruton\u0026rsquo;s tyrosine kinase inhibitors: predictive ability and interpretability. Mol Divers. 2024;28(4):2429\u0026ndash;47.\u003c/li\u003e\n\u003cli\u003eJung Y, Hu J. AK-fold averaging cross-validation procedure. J Nonparametric Stat. 2015;27(2):167\u0026ndash;79.\u003c/li\u003e\n\u003cli\u003eBernert RA, Hilberg AM, Melia R, Kim JP, Shah NH, Abnousi F. Artificial Intelligence and Suicide Prevention: A Systematic Review of Machine Learning Investigations. Int J Environ Res Public Health. 2020 Aug 15;17(16):5929.\u003c/li\u003e\n\u003cli\u003eReese BM, Chen MS, Nekkanti M, Mulawa MI. Prevalence and Risk Factors of Women\u0026rsquo;s Past-Year Physical IPV Perpetration and Victimization in Tanzania. J Interpers Violence. 2021 Feb;36(3\u0026ndash;4):1141\u0026ndash;67.\u003c/li\u003e\n\u003cli\u003eChikhungu LC, Amos M, Kandala N, Palikadavath S. Married Women\u0026rsquo;s Experience of Domestic Violence in Malawi: New Evidence From a Cluster and Multinomial Logistic Regression Analysis. J Interpers Violence. 2021 Sep;36(17\u0026ndash;18):8693\u0026ndash;714.\u003c/li\u003e\n\u003cli\u003eLasong J, Zhang Y, Muyayalo KP, Njiri OA, Gebremedhin SA, Abaidoo CS, et al. Domestic violence among married women of reproductive age in Zimbabwe: a cross sectional study. BMC Public Health. 2020 Dec;20(1):354.\u003c/li\u003e\n\u003cli\u003eHossain Md, Asadullah Md, Rahaman A, Miah Md, Hasan M, Paul T, et al. Prediction on Domestic Violence in Bangladesh during the COVID-19 Outbreak Using Machine Learning Methods. Appl Syst Innov. 2021 Oct 13;4(4):77.\u003c/li\u003e\n\u003cli\u003eRahman R, Khan MdNA, Sara SS, Rahman MdA, Khan ZI. A comparative study of machine learning algorithms for predicting domestic violence vulnerability in Liberian women. BMC Womens Health. 2023 Oct 17;23(1):542.\u003c/li\u003e\n\u003cli\u003eIkekwuibe IC, Okoror CEM. The pattern and socio-cultural determinants of intimate partner violence in a Nigerian rural community. Afr J Prim Health Care Fam Med [Internet]. 2021 Jun 1 [cited 2025 Apr 28];13(1). Available from: https://phcfm.org/index.php/phcfm/article/view/2435\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"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":"bmc-public-health","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"pubh","sideBox":"Learn more about [BMC Public Health](http://bmcpublichealth.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/pubh/default.aspx","title":"BMC Public Health","twitterHandle":"@BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Machine Learning, Random Forest Algorithm, Intimate Partner Violence, Predictive Model, Rwanda","lastPublishedDoi":"10.21203/rs.3.rs-7686865/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7686865/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e\u003cp\u003eIntimate partner violence (IPV) is a serious global public health issue that predominantly affects women. Despite its widespread nature, IPV data and prevention strategies remain limited, particularly in low- and middle-income countries (LMICs), with sub-Saharan Africa bearing a significant burden. This study seeks to develop and validate a robust machine learning model to predict IPV and its predictors in Rwanda.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e\u003cp\u003eThe study used five different machine learning algorithms, including the Least Absolute Shrinkage and Selection Operator (LASSO), logistic regression (LR), Random Forest, Support Vector Machine, and Classification and Regression Trees (CART) to develop models using data of 14,634 participants from the 2019\u0026ndash;2020 Demographic and Health Survey. We used multiple imputation for missing data, and the Shapley Additive Explanations (SHAP) technique was used to select the most predictive variables. The data was randomly split into 80% and 20% for training and testing, respectively. To achieve a balanced dataset, the Synthetic Minority Over-sampling Technique was used on the training subset. The algorithms were evaluated using F1 score, AUC, accuracy, precision, sensitivity, and specificity.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e\u003cp\u003eThe random forest algorithm (accuracy and Receiver Operating Characteristic (ROC) curve of 99.1% and 100%, respectively) proved remarkably successful in accurately predicting IPV, outperforming other algorithms. Interestingly, the model\u0026rsquo;s performance revealed a significant improvement when utilizing only fifteen variables to predict IPV. The algorithm owes its success to its remarkable capability to identify crucial predictor features, with the top four being arguing with a partner, jealous partner, burning food, and insisting on knowing the respondent\u0026rsquo;s whereabouts.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e\u003cp\u003eThe random forest algorithm was a highly effective tool for identifying potential predictors of IPV. Machine learning models may serve as an effective approach to address IPV and could help in early intervention and prevention by strengthening policy enforcement and integrating predictive tools into healthcare systems.\u003c/p\u003e","manuscriptTitle":"Application of Machine Learning in predicting intimate partner-based violence among reproductive age women in Rwanda","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-11-25 14:43:30","doi":"10.21203/rs.3.rs-7686865/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"editorInvitedReview","content":"","date":"2025-11-15T10:45:00+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"270339238680879995165094350838578689918","date":"2025-11-13T13:10:30+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-11-13T08:27:32+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-09-25T08:10:02+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-09-25T07:21:51+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-09-25T07:19:36+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Public Health","date":"2025-09-22T16:37:53+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-public-health","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"pubh","sideBox":"Learn more about [BMC Public Health](http://bmcpublichealth.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/pubh/default.aspx","title":"BMC Public Health","twitterHandle":"@BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"03cae97a-2acc-4679-a864-33f7f82873cb","owner":[],"postedDate":"November 25th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2025-11-25T14:43:30+00:00","versionOfRecord":[],"versionCreatedAt":"2025-11-25 14:43:30","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7686865","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7686865","identity":"rs-7686865","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}