Predicting Perinatal Mortality Based on Maternal Health Status and Health Insurance Service using Homogeneous Ensemble Machine Learning Methods

Predicting Perinatal Mortality Based on Maternal Health Status and Health Insurance Service using Homogeneous Ensemble Machine Learning Methods | 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 Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Predicting Perinatal Mortality Based on Maternal Health Status and Health Insurance Service using Homogeneous Ensemble Machine Learning Methods Dawit S Bogale, Tesfamariam M Abuhay, Belayneh E Dejene This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-1445740/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 28 Dec, 2022 Read the published version in BMC Medical Informatics and Decision Making → Version 1 posted 5 You are reading this latest preprint version Abstract Background Perinatal mortality in Ethiopia is the highest in Africa, with 68 per 1000 pregnancies intrapartum deaths (death during the delivery). It is mainly associated with home delivery, which contributes for more than 75% of perinatal deaths. Financial constraints have a significant impact on timely access to maternal health (MH) care. Financial incentives, such as health insurance, can address the demand- and supply-side factors. This study, hence, aims to predict perinatal mortality based on maternal health status and health insurance service using homogeneous ensemble machine learning methods Methods The data was collected from Ethiopian demographic health survey from 2011 to 2019 G.C. The data were pre-processed to get quality data that are suitable for a homogenous ensemble machine-learning algorithm to develop a model that predicts perinatal mortality. Results For constructing the proposed model, three experiments were conducted using random forest, gradient boosting, and cat boost algorithms. The overall accuracy of random forest, gradient boosting, and cat boost with 17 features is 89.95%, 90.24%, and 82%, respectively. Conclusions We finally concluded that perinatal mortality over time in Ethiopia is decreasing. We found out that perinatal mortality in Ethiopia is associated with risk factors such as community-based health insurance, mother's educational level, residence, mother age, wealth status, distance to the health facility, preterm, smoke cigarette, anemia level, haemoglobin level, and marital status. homogenous ensembles machine learning perinatal mortality maternal health health insurance. Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 1. Background Perinatal mortality refers to a fatal death at or after 28 weeks of pregnancy (stillbirth) and includes death within 7 days of life after birth [ 1 ][ 2 ]. According to the World Health Organization (WHO) 2019 report, there were 2.6 million newborn infants globally, but more than 8200 died within a day [ 3 ]. Among the 133 million newborn infants alive each year, 2.8 million died in the first week of life after birth/at birth, and the majority occurred in low-income level countries [ 3 ]. Given the reaching deadlines for reaching the Millennium Development Goals, the international community supports low- and middle-level income countries to renew their commitment to reducing maternal and infants mortality rates by improving access to maternal, neonatal, and perinatal health services [ 4 ]. Perinatal mortality in Ethiopia is the highest in Africa, with 68 per 1000 pregnancies Intrapartum deaths (death during the delivery) [ 8 ]. Ethiopia shared and valued the Sustainable Development Goals (SDGs) and has been trying to achieve the target of reducing neonatal mortality to below 12 per 1000 live births, by 2030 [ 9 ]. However, reduction of neonatal, infant and under-five mortalities cannot be realized without substantial reduction of perinatal mortality [ 10 ]. It is mostly associated with home deliveries, which contributed for more than 75% of all perinatal deaths due to the lack of awareness about health insurance services during birth, and it continued to be an essential part of the third sustainable development goal which aims to end preventable children's deaths by 2030 [ 9 ]. Financial constraints have a significant impact on timely access to maternal health (MH) care, such as Antenatal Care (ANC), skilled care at delivery, access to facility-based deliveries, postnatal care (PNC), and perinatal [ 7 ]. Over 100 million individuals pay out-of-pocket (OOP) payments to get health treatments that have proven difficult to obtain for millions of poor people, resulting in increased morbidity and mortality [ 5 ]. WHO recommends community-based health insurance (CBHI) as one of the approaches for reducing OOP expenditures for registered families which, in turn, reduce morbidity and mortality [ 6 ]. The association of CBHI with reduced maternal and infant mortality was apparent but it is impossible to reduce the infant mortality rate, without reducing the perinatal mortality [ 7 ]. Financial incentives, such as health insurance, can address the demand- and supply factors that may possibly impacting maternal, neonatal, and perinatal health results [ 11 ]. To this end, the Ethiopian Ministry of Health has been working for years to make health services accessible for women through community and facility-based interventions to increase survival of newborn and children [ 9 ]. Despite these interventions, perinatal death remains an issue in Ethiopia, in particular; home delivery remains the challenge to reduce perinatal mortality [ 11 ]. Still, 74% of women give birth outside health institutions without skilled care attendants in Ethiopia [ 8 ][ 12 ][ 13 ]. This study, hence, aims at predicting perinatal mortality based on maternal health status and health insurance service using homogeneous ensemble machine learning methods by investigating the following research questions (1) what is the underline structure and evolution of perinatal mortality in Ethiopia over time? (2) Which homogeneous ensemble machine learning methods is suitable to predict perinatal mortality in Ethiopia effectively? (3) What are the determinant factors of perinatal mortality in Ethiopia? (4) What are the important rules that may shape strategies, policies and interventions towards preventing and/or reducing perinatal mortality in Ethiopia? The rest of this paper is organized as follows: Section 2 presents related works, Section 3 discusses materials and methods used, Section 4 mentions experimental setup and result discussion, and Section 5 presents conclusion. 2. Related Work Several studies investigated perinatal mortality in Ethiopia using different methods. Getachew et al. [ 14 ] investigated perinatal mortality and associated risk factors using a case-control study between 2008 and 2010 using a total of 1356 newborns’ data (452 cases and 904 controls). Subgroup binary logistic regression analyses were done to identify associated risk factors for perinatal mortality, stillbirths, and early neonatal deaths. The study reported that the perinatal mortality rate was 85/1000, and after or at 28 weeks of birth death accounts for 87% [ 14 ]. Adjusted odds ratios revealed that obstructed labor, malpresentation, preterm birth, death during the delivery haemorrhage, and hypertensive disorders of pregnancy was an independent predictor for high perinatal mortality. Another study was conducted by Yemisrach et al. [ 15 ] on factors associated with perinatal mortality among public health deliveries in Addis Ababa, Ethiopia using an unmatched case-control study and secondary data that was collected between 1st January up to 30th February 2015. In this study, a total of 1113 (376 cases and 737 controls) maternal charts were reviewed and the mean age of the mothers for cases and controls were 26.47 ± 4.87 and 26.95 ± 4.68, respectively. Five hundred ninety-seven (53.6%) mothers delivered for the first time and factors that are significantly associated with increased risk of perinatal mortality were birth interval less than 2 years, preterm delivery, anemia, congenital anomaly, previous history of early neonatal death, and low birth weight. Use of partograph was also associated with decreased risk of perinatal mortality. Bekele et al. [ 16 ] studied the effect of community-based health insurance on utilization of outpatient health care services in Yirgalem town, Southern Ethiopia. This study used both quantitative and qualitative (mixed) approaches using a comparative cross-sectional study design. Randomly selected sample of 405 (135 members and 270 non-members) household heads were used for quantitative analysis. Multivariate logistic regression was employed to identify the effect of community-based health insurance on healthcare utilization. This study reveals that members of households with community-based health insurance were about three times more likely to utilize outpatient care than their non-member counterparts [AOR: 2931; 95% CI (1.039, 7.929); p-value = 0.042]. Finally, the researchers conclude that community-based health insurance is an effective tool to increase the utilization of healthcare services and provide the scheme to member households. However, the aforementioned studies focused on identifying determinant risk factors only. Besides, these studies did not develop a predictive model, did not design an artefact that can be used by potential users, and did not generate rules that allow the development of evidence-based preventive strategies, policies and interventions. On the other hand, machine learning algorithms have proven their effectiveness and efficiency in predicting child mortality in African countries such as South Africa [ 17 ] and Uganda [ 18 ]. This study, hence, motivated to fill these gaps by identifying risk factors, constructing a predictive model, design artifact, and generate rules that help to develop evidence-based policies and interventions towards perinatal mortality in Ethiopia. 3. Materials And Methods Figure 1 depicts the proposed model architecture that was implemented in this study to construct a predictive model, identify risk factors, extract relevant rules, and design artifacts. 3.1. DATA COLLECTION In this study we used secondary data, the Ethiopia Demographic and Health Surveys (EDHS) which was collected by the Ethiopian Central Statistical Agency in 2011, 2016, and 2019 G.C, in five years intervals. The EDHSs are nationally representative household surveys that collect data for a variety of demographic, health, and nutrition monitoring and impact evaluation purposes. 3.2. DATA PREPROCESSING The raw data contains 45 columns and 109531 instances. Data imputation (mode for categorical data and mean for continuous data) method was employed to substitute the missing values. Outliers were identified using a boxplot and replaced using the Interquartile Range (IQR) scores. Binning data discretization was applied to transform some of the features. For example, the feature ‘education level of mothers (v106)’ has 8 different values which were transformed into five different values (illiterate (1), grade 1–8 (elementary), grade9-12 (secondary), grade 12+ (tertiary), and higher education (university and college)). The synthetic minority over-sampling technique (SMOTE) was implemented to handle the class imbalance in the training dataset. The main reason that we use SMOTE is it avoids loss of valuable information [ 22 ][ 23 ]. Then, four experiments were conducted using filter and wrapper methods to select the relevant features for developing a perinatal mortality prediction model. As a result, the sequential backward feature selection method has registered the highest performance with 90.5% of accuracy and produced 13 important features. Besides, the domain expert’s recommended additional 4 features and the total features selected for further analysis are 17, see Table 1 . Table 1 Features selected by sequential forward feature selection No Feature code Feature description 1. Bord Birth interval 2. V024 Region 3. V013 Maternal age 4. V190 Wealth index 5. V717 Maternal occupation 6. V457 Anemia level 7. V394 Visited health facility last 12 week 8. V501 Marital status 9. V312 Current contraceptive 10. V161 Types of cooking fuel 11. V106 Educational level 12. V228 Preterm 13. V455 Hemoglobin level 14. V025 Place of residence 15. V481a Community/mutual health insurance 16. V463a Smoke cigarettes 17. V463c Chews tobacco 4. Experimental Setup And Results In Discussion 4.1. What is the underline structure and evolution of perinatal mortality in Ethiopia over time? The perinatal mortality has been reducing over time in Ethiopia. This is because of increase in the number of hospitals, especially in rural areas and the introduction of community-based health insurance which encourages pregnant women to visit hospital to give birth. But, due to COVID 19 pandemic, the data collected in 2019 were twice smaller than previous years, as shown in Fig. 2 . The perinatal mortality across the regions (Tigray, afar, Amhara, Oromia, Somali, Benishangul, SNNP, Gambella, Harari, Addis Ababa, and Dire Dawa) of Ethiopia was also investigated. Among nine regions and two administrative cities of Ethiopia, Amhara and Oromia regions registered higher perinatal mortality compared to other regions, as shown in Fig. 3 . Having health insurance service leads to almost zero perinatal mortality because it helps mothers to get access to health facility in time, as shown in Fig. 4 . 4.2. Which homogeneous ensemble machine learning algorithm predict perinatal mortality in Ethiopia effectively? Three experiments were conducted to build a perinatal mortality predictive model using classification algorithms namely: Gradient Boost, CatBoost, and random forest classifiers. Grid search was applied to tune the hyperparameters of these algorithms. As a result, gradient boosting (with parameters: criterion='entropy', max_depth = 15, max_Depth = max_Depth, bootstrap = True, N_estimators = warn, N_jobs = none, random state = 42) performed better with 99.72% recall, 90.24% accuracy, 92.80% f1-score, 86.96% ROC and 87.24% precision. The recall indicates that there is a maximized true positive rate and a minimized false-negative rate meaning; there is a minimum false-negative rate. The confusion matrix of Gradient Boosting algorithms is presented in Table 2 . Table 2 Confusion matrix of gradient boosting Predicted Class Died Alive Actual class Died 8125 2736 Alive 167 18704 Therefore, the gradient boost algorithm is selected as the best homogenous ensemble machine learning algorithm for predicting perinatal mortality based on maternal health status and health insurance service in the study area. The overall results of each experiment are summarized in Table 3 . Table 3 Overall performance of models Evaluation Algorithms Gradient Boost (%) Cat Boost (%) Random forest (%) Accuracy 90.24 81.45 89.95 Precision 87.24 82.01 86.42 Recall 99.72 90.75 99.54 ROC 86.96 77.98 86.50 F1_Score 92.80 86.16 92.72 4.3. What are the determinant factors of perinatal mortality in Ethiopia? Feature importance analysis was conducted to identify determinant risk factors of perinatal mortality in Ethiopia using the best performing model which was developed using gradient boosting. As a result, factors that are significantly associated with increased risk of perinatal mortality are birth interval less than 2 years, preterm delivery, anemia, congenital anomaly, educational status, family size, occupation, marital status, traveling time to the nearest health institution, perceived quality of care, the first choice of place for treatment during illness and expected healthcare cost of recent treatment, prematurity, low birth weight, previous history of perinatal death, not receiving tetanus toxoid immunization, and lack of iron supplementation, see Table 4 . Table 4 Risk factors with feature importance No Feature code Feature description Feature importance value 1. Bord Birth interval 0.291119 2. V024 Region 0.122834 3. V013 Maternal age 0.077887 4. V190 Wealth index 0.072292 5. V717 Maternal occupation 0.055206 6. V457 Anemia level 0.054080 7. V394 Visited health facility last 12 week 0.038728 8. V501 Marital status 0.030207 9. V312 Current contraceptive 0.026625 10. V161 Types of cooking fuel 0.026059 11. V106 Educational level 0.021210 12. V228 Preterm 0.019435 13. V455 Hemoglobin level 0.016383 14. V025 Place of residence 0.009877 15. V481a Community/mutual health insurance 0.007053 16. V463a Smoke cigarettes 0.006037 17. V463c Chews tobacco 0.002598 4.4 What are the important rules that may shape strategies, policies and interventions towards reducing and/or preventing perinatal mortality in Ethiopia? The most relevant rules were generated from the best-performed algorithm (gradient boost) model, and the rules were validated by the domain experts. Sample rules are presented here below and Fig. 5 presents decision tree of relevant rule that were generated by the best performing algorithm: Rule1:- if currently breast feeding and preterm == 'no' AND maternal education== 'no education' AND wanted least children == 'wanted then' AND smoke ciggrate == 'no' AND health insurance provide by employer == 'no' AND smoke Tabaco == 'never in union' AND types of cooking fuel == 'wood' AND occupation == 'not working AND wealth index== 'poorest' AND maternal age == '35–39' AND place of residence == 'rural' AND Community based health insurance == 'no' AND Then child=='Alive' Rule 2:- if currently breast feeding and preterm == 'no' AND maternal education== 'no education' AND wanted least children == 'wanted then' AND smoke ciggrate == 'no' AND health insurance provide by employer == 'no' AND smoke Tabaco == 'never in union' AND types of cooking fuel == 'wood' AND occupation == 'not working AND wealth index== 'poorest' AND maternal age == '40–44' AND place of residence == 'urban' AND Community based health insurance == 'no' AND Then child=='Died' Rule3:- if currently breast feeding and preterm == 'no' AND maternal education== 'no education' AND wanted least children == 'wanted then' AND smoke ciggrate == 'no' AND health insurance provide by employer == 'no' AND smoke Tabaco == 'never in union' AND types of cooking fuel == 'wood' AND occupation == 'not working AND wealth index== 'poorest' AND maternal age == '45–49' AND place of residence == 'rural' AND Community based health insurance == 'no' AND Then children=='Alive' Rule4:- if currently breast feeding and preterm == 'no' AND maternal education== 'no education' AND wanted least children == 'wanted then' AND smoke ciggrate == 'no' AND health insurance provide by employer == 'no' AND smoke Tabaco == 'never in union' AND types of cooking fuel == 'wood' AND occupation == 'not working AND wealth index== 'poorest' AND maternal age == '45–49' AND place of residence == 'rural' AND Community based health insurance == 'no' AND Then children=='Alive' Rule 5:- if currently breast feeding and preterm == 'no' AND maternal education== 'no education' AND wanted least children == 'wanted then' AND smoke ciggrate == 'no' AND health insurance provide by employer == 'no' AND smoke Tabaco == 'never in union' AND types of cooking fuel == 'wood' AND occupation == 'not working AND wealth index== 'poorest' AND maternal age == '15–19' AND place of residence == 'rural' AND Community based health insurance == 'no' AND Then children== 'Died' 5. Discussion As we discussed in risk factors identification section, the risk factors were identified using feature importance techniques and rules were generated using best performing algorithm which is gradient boosting as show in the experimental section. As we have discussed in the experimental result section, the proposed system achieved an overall performance of 90.24%, which is better a result compared to a result achieved by previous studies using gradient boosting machine learning algorithm which was 83% overall performance. We have deployed the model on cloud using Heroku and Flask framework and can be freely accessed via this link: http://perinatal-mortality.herokuapp.com/ 6. Conclusion This study aims at developing a model that predicts perinatal mortality in the case of Ethiopia by using homogeneous ensemble machine learning methods. The gradient boost algorithm has registered the highest performance with 99.72% recall, 90.24% accuracy, 92.80% f1-score, 86.96% ROC and 87.24% precision. We identified the determinant risk factors of perinatal mortality with feature importance techniques such as maternal residence, level of education, birth interval, and community-based health insurance. The most relevant rules, that helps to formulate evidence-based strategies and policies towards maintaining perinatal mortality, were generated from the best performing model, and the rules were validated by the domain experts. Abbreviations ANC Antenatal Care CBHI Community-Based Health Insurance EDHS Ethiopia Demographic Health Survey MH Maternal Health OOP Out-Of-Pocket PNC Postnatal Care ROC Receiver Operating Characteristics SDGs Sustainable Development Goals SMOTE Synthetic Minority Over-Sampling Technique WHO World Health Organization Declarations Acknowledgements We would like to acknowledge the Ethiopia central statistical agency for providing us the data. Funding The research was supported by the University of Gondar research and community service vice president's office. Availability of data and materials The datasets generated and/or analysed during the current study are available in the ‘perinatal_dataset-’ repository, https://github.com/dawitemu1/perinatal_dataset-. Ethics approval and consent to participate All methods used in this study followed guidelines and regulationsthatwere approved by the institutional reviewboardof the University of Gondar. Members of the board are Professor FelekeMoges, Mr. NiguseYigzaw, Mr. AbiyotEndale, Dr. MisayeMulate, Dr. AlemayehuTekelu and Dr. BimerewAdmasu. Consent for publication Not applicable. Competing interests The authors report that they have no conflicts. References [1] A. B. Comfort, L. A. Peterson, and L. E. Hatt, “Effect of health insurance on the use and provision of maternal health services and maternal and neonatal health outcomes: A systematic review,” J. Heal. Popul. Nutr., vol. 31, no. 4 SUPPL.2, 2013, doi: 10.3329/jhpn.v31i4.2361. [2] V. Jain and J. M. Chatterjee, “Machine Learning with Health Care Perspective: Machine Learning and Healthcare,” no. March, 2020, doi: 10.1007/978-3-030-40850-3. [3] R. Rasaily et al., “Effect of home-based newborn care on neonatal and infant mortality: A cluster randomised trial in India,” BMJ Glob. Heal., vol. 5, no. 9, pp. 1–11, 2020, doi: 10.1136/bmjgh-2017-000680. [4] J. R. Daw, T. N. A. Winkelman, V. K. Dalton, K. B. Kozhimannil, and L. K. Admon, “Medicaid expansion improved perinatal insurance continuity for low-income women,” Health Aff., vol. 39, no. 9, pp. 1531–1539, 2020, doi: 10.1377/hlthaff.2019.01835. [5] E. Jukes, Encyclopedia of Machine Learning and Data Mining (2nd edition), vol. 32, no. 7/8. 2018. [6] W. Soors, N. Devadasan, V. Durairaj, and B. Criel, “Community Health Insurance and Universal Coverage: Multiple paths, many rivers to cross,” pp. 1–122, 2010. [7] N. Haven et al., “Community-based health insurance increased health care utilization and reduced mortality in children under-5, around Bwindi Community Hospital, Uganda between 2015 and 2017,” Front. Public Heal., vol. 6, no. OCT, 2018, doi: 10.3389/fpubh.2018.00281. [8] B. J. Akombi and A. M. Renzaho, “Perinatal mortality in sub-saharan africa: A meta-analysis of demographic and health surveys,” Ann. Glob. Heal., vol. 85, no. 1, pp. 1–8, 2019, doi: 10.5334/aogh.2348. [9] P. R. Ghimire, K. E. Agho, A. M. N. Renzaho, M. K. Nisha, M. Dibley, and C. Raynes-Greenow, “Factors associated with perinatal mortality in Nepal: Evidence from Nepal demographic and health survey 2001-2016,” BMC Pregnancy Childbirth, vol. 19, no. 1, pp. 1–12, 2019, doi: 10.1186/s12884-019-2234-6. [10] Z. A. Hassan and M. J. Ahmed, “Factors associated with immunisation coverage of children aged 12- 24 months in Erbil / Iraq 2017-2018,” vol. 24, no. 08, pp. 12222–12235, 2020, doi: 10.37200/IJPR/V24I8/PR281205. [11] R. A. Knuppel and J. H. Shepherd, “Perinatal mortality rates,” Br. Med. J., vol. 280, no. 6228, p. 1376, 1980, doi: 10.1136/bmj.280.6228.1376. [12] B. H. Jena, G. A. Biks, K. A. Gelaye, and Y. K. Gete, “Magnitude and trend of perinatal mortality and its relationship with inter-pregnancy interval in Ethiopia: A systematic review and meta-analysis,” BMC Pregnancy Childbirth, vol. 20, no. 1, pp. 1–13, 2020, doi: 10.1186/s12884-020-03089-2. [13] G. T. Debelew, “Magnitude and Determinants of Perinatal Mortality in Southwest Ethiopia,” J. Pregnancy, vol. 2020, 2020, doi: 10.1155/2020/6859157. [14] G. Bayou and Y. Berhan, “Perinatal mortality and associated risk factors: a case control study.,” Ethiop. J. Health Sci., vol. 22, no. 3, pp. 153–62, 2012. [15] Y. Getiye and M. Fantahun, “Factors associated with perinatal mortality among public health deliveries in Addis Ababa, Ethiopia, an unmatched case control study,” BMC Pregnancy Childbirth, vol. 17, no. 1, pp. 1–7, 2017, doi: 10.1186/s12884-017-1420-7. [16] B. Demissie and K. G. Negeri, “Effect of community-based health insurance on utilization of outpatient health care services in southern ethiopia: A comparative cross-sectional study,” Risk Manag. Healthc. Policy, vol. 13, pp. 141–153, 2020, doi: 10.2147/RMHP.S215836. [17] C. Kabudula et al., “Evaluation of machine learning methods for predicting the risk of child mortality in South Africa,” 2019. [18] G. Nguyen, “Evaluating statistical and machine learning methods to predict risk of in-hospital child mortality in Uganda,” 2016. [19] D. D. Atnafu, H. Tilahun, and Y. M. Alemu, “Community-based health insurance and healthcare service utilisation, North-West, Ethiopia: A comparative, cross-sectional study,” BMJ Open, vol. 8, no. 8, pp. 1–6, 2018, doi: 10.1136/bmjopen-2017-019613. [20] K. Shiferaw, B. Mengiste, T. Gobena, and M. Dheresa, “The effect of antenatal care on perinatal outcomes in Ethiopia: A systematic review and meta-analysis,” PLoS One, vol. 16, no. 1 January, pp. 1–19, 2021, doi: 10.1371/journal.pone.0245003. [21] D. Prasad et al., “The effect of community-based health insurance on the utilization of modern health care services : Evidence from Burkina Faso,” vol. 90, pp. 214–222, 2009, doi: 10.1016/j.healthpol.2008.09.015. [22] I. Journal and C. Science, “Class Imbalance Problem in Data Mining : Review,” vol. 2, no. 1, 2013. [23] R. P. Ribeiro, “SMOTE for Regression,” no. October 2015, 2013, doi: 10.1007/978-3-642-40669-0. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 28 Dec, 2022 Read the published version in BMC Medical Informatics and Decision Making → Version 1 posted Editorial decision: Major revision 02 Jun, 2022 Editor assigned by journal 31 May, 2022 Editor invited by journal 21 Mar, 2022 Submission checks completed at journal 21 Mar, 2022 First submitted to journal 12 Mar, 2022 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 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-1445740","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":92374592,"identity":"69763d0d-1cda-4732-82b6-0c25ebadf624","order_by":0,"name":"Dawit S Bogale","email":"","orcid":"","institution":"University of Gondar","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Dawit","middleName":"S","lastName":"Bogale","suffix":""},{"id":92374593,"identity":"94e9bc27-411b-48e0-abd4-3d47d93715f5","order_by":1,"name":"Tesfamariam M Abuhay","email":"data:image/png;base64,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","orcid":"","institution":"University of Gondar","correspondingAuthor":true,"submittingAuthor":false,"prefix":"","firstName":"Tesfamariam","middleName":"M","lastName":"Abuhay","suffix":""},{"id":92374594,"identity":"73948b0c-1470-436d-a3ad-aba2f11c77cb","order_by":2,"name":"Belayneh E Dejene","email":"","orcid":"","institution":"University of Gondar","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Belayneh","middleName":"E","lastName":"Dejene","suffix":""}],"badges":[],"createdAt":"2022-03-12 20:59:05","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-1445740/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-1445740/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s12911-022-02084-1","type":"published","date":"2022-12-28T18:07:21+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":19489159,"identity":"73635f03-8676-44f7-a8d8-8fbfd3c75d66","added_by":"auto","created_at":"2022-03-22 16:22:29","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":97655,"visible":true,"origin":"","legend":"\u003cp\u003eThe proposed model architecture\u0026nbsp;\u003c/p\u003e","description":"","filename":"FIG1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-1445740/v1/587fe2b3d2278131756dafcb.jpg"},{"id":19489282,"identity":"18f64512-71d1-4222-b46f-0c03e21657b9","added_by":"auto","created_at":"2022-03-22 16:25:29","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":22797,"visible":true,"origin":"","legend":"\u003cp\u003ePerinatal mortality over time in Ethiopia\u003c/p\u003e","description":"","filename":"FIG2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-1445740/v1/61f12ed91d9c8de5017f4c0f.jpg"},{"id":19489160,"identity":"719ab4c8-8048-4c2c-9b84-8ce1d1939f86","added_by":"auto","created_at":"2022-03-22 16:22:29","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":43706,"visible":true,"origin":"","legend":"\u003cp\u003eThe Perinatal Mortality across the Regions in Ethiopia\u003c/p\u003e","description":"","filename":"FIG3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-1445740/v1/e89ed55c83ffdc38d2daf1b0.jpg"},{"id":19489561,"identity":"7ef7221b-796a-4c5c-8523-19f1662a1fd0","added_by":"auto","created_at":"2022-03-22 16:28:29","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":25092,"visible":true,"origin":"","legend":"\u003cp\u003ePerinatal Mortality Compared to Health Insurance\u0026nbsp;\u003c/p\u003e","description":"","filename":"FIG4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-1445740/v1/5ddd7b893e00f9c7ccc81e94.jpg"},{"id":19489163,"identity":"689f6864-dc0e-41a3-965c-95073cf4a38a","added_by":"auto","created_at":"2022-03-22 16:22:29","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":104006,"visible":true,"origin":"","legend":"\u003cp\u003egenerated rule by decision tree\u0026nbsp;\u003c/p\u003e","description":"","filename":"FIG5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-1445740/v1/f0b3483dbacafa670d48547c.jpg"},{"id":44715623,"identity":"7fbf5d84-111a-4c9e-8102-82f1c6f6694c","added_by":"auto","created_at":"2023-10-16 18:14:44","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":631486,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-1445740/v1/0613290b-34a5-4883-96b0-e161fb015864.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Predicting Perinatal Mortality Based on Maternal Health Status and Health Insurance Service using Homogeneous Ensemble Machine Learning Methods","fulltext":[{"header":"1. Background","content":"\u003cp\u003ePerinatal mortality refers to a fatal death at or after 28 weeks of pregnancy (stillbirth) and includes death within 7 days of life after birth [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e][\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. According to the World Health Organization (WHO) 2019 report, there were 2.6\u0026nbsp;million newborn infants globally, but more than 8200 died within a day [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Among the 133\u0026nbsp;million newborn infants alive each year, 2.8\u0026nbsp;million died in the first week of life after birth/at birth, and the majority occurred in low-income level countries [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Given the reaching deadlines for reaching the Millennium Development Goals, the international community supports low- and middle-level income countries to renew their commitment to reducing maternal and infants mortality rates by improving access to maternal, neonatal, and perinatal health services [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e].\u003c/p\u003e \u003cp\u003ePerinatal mortality in Ethiopia is the highest in Africa, with 68 per 1000 pregnancies Intrapartum deaths (death during the delivery) [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. Ethiopia shared and valued the Sustainable Development Goals (SDGs) and has been trying to achieve the target of reducing neonatal mortality to below 12 per 1000 live births, by 2030 [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. However, reduction of neonatal, infant and under-five mortalities cannot be realized without substantial reduction of perinatal mortality [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. It is mostly associated with home deliveries, which contributed for more than 75% of all perinatal deaths due to the lack of awareness about health insurance services during birth, and it continued to be an essential part of the third sustainable development goal which aims to end preventable children's deaths by 2030 [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eFinancial constraints have a significant impact on timely access to maternal health (MH) care, such as Antenatal Care (ANC), skilled care at delivery, access to facility-based deliveries, postnatal care (PNC), and perinatal [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. Over 100\u0026nbsp;million individuals pay out-of-pocket (OOP) payments to get health treatments that have proven difficult to obtain for millions of poor people, resulting in increased morbidity and mortality [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. WHO recommends community-based health insurance (CBHI) as one of the approaches for reducing OOP expenditures for registered families which, in turn, reduce morbidity and mortality [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. The association of CBHI with reduced maternal and infant mortality was apparent but it is impossible to reduce the infant mortality rate, without reducing the perinatal mortality [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. Financial incentives, such as health insurance, can address the demand- and supply factors that may possibly impacting maternal, neonatal, and perinatal health results [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. To this end, the Ethiopian Ministry of Health has been working for years to make health services accessible for women through community and facility-based interventions to increase survival of newborn and children [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. Despite these interventions, perinatal death remains an issue in Ethiopia, in particular; home delivery remains the challenge to reduce perinatal mortality [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. Still, 74% of women give birth outside health institutions without skilled care attendants in Ethiopia [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e][\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e][\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThis study, hence, aims at predicting perinatal mortality based on maternal health status and health insurance service using homogeneous ensemble machine learning methods by investigating the following research questions (1) what is the underline structure and evolution of perinatal mortality in Ethiopia over time? (2) Which homogeneous ensemble machine learning methods is suitable to predict perinatal mortality in Ethiopia effectively? (3) What are the determinant factors of perinatal mortality in Ethiopia? (4) What are the important rules that may shape strategies, policies and interventions towards preventing and/or reducing perinatal mortality in Ethiopia?\u003c/p\u003e \u003cp\u003eThe rest of this paper is organized as follows: Section 2 presents related works, Section 3 discusses materials and methods used, Section 4 mentions experimental setup and result discussion, and Section 5 presents conclusion.\u003c/p\u003e"},{"header":"2. Related Work","content":"\u003cp\u003eSeveral studies investigated perinatal mortality in Ethiopia using different methods. Getachew et al. [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e] investigated perinatal mortality and associated risk factors using a case-control study between 2008 and 2010 using a total of 1356 newborns\u0026rsquo; data (452 cases and 904 controls). Subgroup binary logistic regression analyses were done to identify associated risk factors for perinatal mortality, stillbirths, and early neonatal deaths. The study reported that the perinatal mortality rate was 85/1000, and after or at 28 weeks of birth death accounts for 87% [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. Adjusted odds ratios revealed that obstructed labor, malpresentation, preterm birth, death during the delivery haemorrhage, and hypertensive disorders of pregnancy was an independent predictor for high perinatal mortality.\u003c/p\u003e \u003cp\u003eAnother study was conducted by Yemisrach et al. [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e] on factors associated with perinatal mortality among public health deliveries in Addis Ababa, Ethiopia using an unmatched case-control study and secondary data that was collected between 1st January up to 30th February 2015. In this study, a total of 1113 (376 cases and 737 controls) maternal charts were reviewed and the mean age of the mothers for cases and controls were 26.47\u0026thinsp;\u0026plusmn;\u0026thinsp;4.87 and 26.95\u0026thinsp;\u0026plusmn;\u0026thinsp;4.68, respectively. Five hundred ninety-seven (53.6%) mothers delivered for the first time and factors that are significantly associated with increased risk of perinatal mortality were birth interval less than 2 years, preterm delivery, anemia, congenital anomaly, previous history of early neonatal death, and low birth weight. Use of partograph was also associated with decreased risk of perinatal mortality. Bekele et al. [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e] studied the effect of community-based health insurance on utilization of outpatient health care services in Yirgalem town, Southern Ethiopia. This study used both quantitative and qualitative (mixed) approaches using a comparative cross-sectional study design. Randomly selected sample of 405 (135 members and 270 non-members) household heads were used for quantitative analysis. Multivariate logistic regression was employed to identify the effect of community-based health insurance on healthcare utilization. This study reveals that members of households with community-based health insurance were about three times more likely to utilize outpatient care than their non-member counterparts [AOR: 2931; 95% CI (1.039, 7.929); p-value\u0026thinsp;=\u0026thinsp;0.042]. Finally, the researchers conclude that community-based health insurance is an effective tool to increase the utilization of healthcare services and provide the scheme to member households.\u003c/p\u003e \u003cp\u003eHowever, the aforementioned studies focused on identifying determinant risk factors only. Besides, these studies did not develop a predictive model, did not design an artefact that can be used by potential users, and did not generate rules that allow the development of evidence-based preventive strategies, policies and interventions. On the other hand, machine learning algorithms have proven their effectiveness and efficiency in predicting child mortality in African countries such as South Africa [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e] and Uganda [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. This study, hence, motivated to fill these gaps by identifying risk factors, constructing a predictive model, design artifact, and generate rules that help to develop evidence-based policies and interventions towards perinatal mortality in Ethiopia.\u003c/p\u003e"},{"header":"3. Materials And Methods","content":"\u003cp\u003eFigure \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e depicts the proposed model architecture that was implemented in this study to construct a predictive model, identify risk factors, extract relevant rules, and design artifacts.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e3.1. DATA COLLECTION\u003c/h2\u003e \u003cp\u003eIn this study we used secondary data, the Ethiopia Demographic and Health Surveys (EDHS) which was collected by the Ethiopian Central Statistical Agency in 2011, 2016, and 2019 G.C, in five years intervals. The EDHSs are nationally representative household surveys that collect data for a variety of demographic, health, and nutrition monitoring and impact evaluation purposes.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e3.2. DATA PREPROCESSING\u003c/h2\u003e \u003cp\u003eThe raw data contains 45 columns and 109531 instances. Data imputation (mode for categorical data and mean for continuous data) method was employed to substitute the missing values. Outliers were identified using a boxplot and replaced using the Interquartile Range (IQR) scores. Binning data discretization was applied to transform some of the features. For example, the feature \u0026lsquo;education level of mothers (v106)\u0026rsquo; has 8 different values which were transformed into five different values (illiterate (1), grade 1\u0026ndash;8 (elementary), grade9-12 (secondary), grade 12+ (tertiary), and higher education (university and college)). The synthetic minority over-sampling technique (SMOTE) was implemented to handle the class imbalance in the training dataset. The main reason that we use SMOTE is it avoids loss of valuable information [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e][\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. Then, four experiments were conducted using filter and wrapper methods to select the relevant features for developing a perinatal mortality prediction model. As a result, the sequential backward feature selection method has registered the highest performance with 90.5% of accuracy and produced 13 important features. Besides, the domain expert\u0026rsquo;s recommended additional 4 features and the total features selected for further analysis are 17, see Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\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\u003eFeatures selected by sequential forward feature selection\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\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 \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFeature code\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eFeature description\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eBord\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eBirth interval\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eV024\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eRegion\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e3.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eV013\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMaternal age\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e4.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eV190\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eWealth index\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e5.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eV717\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMaternal occupation\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e6.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eV457\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAnemia level\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e7.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eV394\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eVisited health facility last 12 week\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e8.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eV501\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMarital status\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e9.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eV312\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCurrent contraceptive\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e10.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eV161\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eTypes of cooking fuel\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e11.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eV106\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eEducational level\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e12.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eV228\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePreterm\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e13.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eV455\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHemoglobin level\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e14.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eV025\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePlace of residence\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e15.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eV481a\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCommunity/mutual health insurance\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e16.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eV463a\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSmoke cigarettes\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e17.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eV463c\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eChews tobacco\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"4. Experimental Setup And Results In Discussion","content":"\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\n\u003ch2\u003e4.1. What is the underline structure and evolution of perinatal mortality in Ethiopia over time?\u003c/h2\u003e\n\u003cp\u003eThe perinatal mortality has been reducing over time in Ethiopia. This is because of increase in the number of hospitals, especially in rural areas and the introduction of community-based health insurance which encourages pregnant women to visit hospital to give birth. But, due to COVID 19 pandemic, the data collected in 2019 were twice smaller than previous years, as shown in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e\n\u003cp\u003eThe perinatal mortality across the regions (Tigray, afar, Amhara, Oromia, Somali, Benishangul, SNNP, Gambella, Harari, Addis Ababa, and Dire Dawa) of Ethiopia was also investigated. Among nine regions and two administrative cities of Ethiopia, Amhara and Oromia regions registered higher perinatal mortality compared to other regions, as shown in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e.\u003c/p\u003e\n\u003cp\u003eHaving health insurance service leads to almost zero perinatal mortality because it helps mothers to get access to health facility in time, as shown in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\n\u003ch2\u003e4.2. Which homogeneous ensemble machine learning algorithm predict perinatal mortality in Ethiopia effectively?\u003c/h2\u003e\n\u003cp\u003eThree experiments were conducted to build a perinatal mortality predictive model using classification algorithms namely: Gradient Boost, CatBoost, and random forest classifiers. Grid search was applied to tune the hyperparameters of these algorithms. As a result, gradient boosting (with parameters: criterion='entropy', max_depth\u0026thinsp;=\u0026thinsp;15, max_Depth\u0026thinsp;=\u0026thinsp;max_Depth, bootstrap\u0026thinsp;=\u0026thinsp;True, N_estimators\u0026thinsp;=\u0026thinsp;warn, N_jobs\u0026thinsp;=\u0026thinsp;none, random state\u0026thinsp;=\u0026thinsp;42) performed better with 99.72% recall, 90.24% accuracy, 92.80% f1-score, 86.96% ROC and 87.24% precision. The recall indicates that there is a maximized true positive rate and a minimized false-negative rate meaning; there is a minimum false-negative rate. The confusion matrix of Gradient Boosting algorithms is presented in Table\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e\n\u003cdiv class=\"gridtable\"\u003e\n\u003ctable id=\"Tab2\" border=\"1\"\u003e\u003ccaption\u003e\n\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\n\u003cdiv class=\"CaptionContent\"\u003e\n\u003cp\u003eConfusion matrix of gradient boosting\u003c/p\u003e\n\u003c/div\u003e\n\u003c/caption\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd colspan=\"2\" rowspan=\"2\" align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n\u003ctd colspan=\"2\" align=\"left\"\u003e\n\u003cp\u003ePredicted Class\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u003cstrong\u003eDied\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u003cstrong\u003eAlive\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd rowspan=\"2\" align=\"left\"\u003e\n\u003cp\u003eActual class\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u003cstrong\u003eDied\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e8125\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e2736\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e\u003cstrong\u003eAlive\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e167\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e18704\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eTherefore, the gradient boost algorithm is selected as the best homogenous ensemble machine learning algorithm for predicting perinatal mortality based on maternal health status and health insurance service in the study area. The overall results of each experiment are summarized in Table\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e.\u003c/p\u003e\n\u003cdiv class=\"gridtable\"\u003e\n\u003ctable id=\"Tab3\" border=\"1\"\u003e\u003ccaption\u003e\n\u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\n\u003cdiv class=\"CaptionContent\"\u003e\n\u003cp\u003eOverall performance of models\u003c/p\u003e\n\u003c/div\u003e\n\u003c/caption\u003e\n\u003cthead\u003e\n\u003ctr\u003e\n\u003cth rowspan=\"2\" align=\"left\"\u003e\n\u003cp\u003eEvaluation\u003c/p\u003e\n\u003c/th\u003e\n\u003cth colspan=\"3\" align=\"left\"\u003e\n\u003cp\u003eAlgorithms\u003c/p\u003e\n\u003c/th\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eGradient Boost (%)\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eCat Boost (%)\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eRandom forest (%)\u003c/p\u003e\n\u003c/th\u003e\n\u003c/tr\u003e\n\u003c/thead\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eAccuracy\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e\u003cstrong\u003e90.24\u003c/strong\u003e\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e81.45\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e89.95\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ePrecision\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e87.24\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e82.01\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e86.42\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eRecall\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e99.72\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e90.75\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e99.54\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eROC\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e86.96\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e77.98\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e86.50\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eF1_Score\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e92.80\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e86.16\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e92.72\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003c/table\u003e\n\u003c/div\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e\n\u003ch2\u003e4.3. What are the determinant factors of perinatal mortality in Ethiopia?\u003c/h2\u003e\n\u003cp\u003eFeature importance analysis was conducted to identify determinant risk factors of perinatal mortality in Ethiopia using the best performing model which was developed using gradient boosting. As a result, factors that are significantly associated with increased risk of perinatal mortality are birth interval less than 2 years, preterm delivery, anemia, congenital anomaly, educational status, family size, occupation, marital status, traveling time to the nearest health institution, perceived quality of care, the first choice of place for treatment during illness and expected healthcare cost of recent treatment, prematurity, low birth weight, previous history of perinatal death, not receiving tetanus toxoid immunization, and lack of iron supplementation, see Table\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e.\u003c/p\u003e\n\u003ctable id=\"Tab4\" border=\"1\"\u003e\u003ccaption\u003e\n\u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e\n\u003cdiv class=\"CaptionContent\"\u003e\n\u003cp\u003eRisk factors with feature importance\u003c/p\u003e\n\u003c/div\u003e\n\u003c/caption\u003e\n\u003cthead\u003e\n\u003ctr\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eNo\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eFeature code\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eFeature description\u003c/p\u003e\n\u003c/th\u003e\n\u003cth align=\"left\"\u003e\n\u003cp\u003eFeature importance value\u003c/p\u003e\n\u003c/th\u003e\n\u003c/tr\u003e\n\u003c/thead\u003e\n\u003ctbody\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e1.\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eBord\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eBirth interval\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.291119\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e2.\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eV024\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eRegion\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.122834\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e3.\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eV013\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eMaternal age\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.077887\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e4.\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eV190\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eWealth index\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.072292\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e5.\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eV717\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eMaternal occupation\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.055206\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e6.\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eV457\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eAnemia level\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.054080\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e7.\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eV394\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eVisited health facility last 12 week\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.038728\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e8.\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eV501\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eMarital status\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.030207\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e9.\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eV312\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eCurrent contraceptive\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.026625\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e10.\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eV161\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eTypes of cooking fuel\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.026059\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e11.\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eV106\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eEducational level\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.021210\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e12.\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eV228\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ePreterm\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.019435\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e13.\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eV455\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eHemoglobin level\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.016383\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e14.\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eV025\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003ePlace of residence\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.009877\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e15.\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eV481a\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eCommunity/mutual health insurance\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.007053\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e16.\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eV463a\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eSmoke cigarettes\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.006037\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003e17.\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eV463c\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"left\"\u003e\n\u003cp\u003eChews tobacco\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd align=\"char\" char=\".\"\u003e\n\u003cp\u003e0.002598\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003c/table\u003e\n\u003ch3 class=\"gridtable\"\u003e4.4\u0026nbsp;What are the important rules that may shape strategies, policies and interventions towards reducing and/or preventing perinatal mortality in Ethiopia?\u003c/h3\u003e\n\u003cp\u003eThe most relevant rules were generated from the best-performed algorithm (gradient boost) model, and the rules were validated by the domain experts. Sample rules are presented here below and Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003e presents decision tree of relevant rule that were generated by the best performing algorithm:\u003c/p\u003e\n\u003cp\u003eRule1:- if currently breast feeding and preterm == 'no' AND maternal education== 'no education' AND wanted least children == 'wanted then' AND smoke ciggrate == 'no' AND health insurance provide by employer == 'no' AND smoke Tabaco == 'never in union' AND types of cooking fuel == 'wood' AND occupation == 'not working AND wealth index== 'poorest' AND maternal age == '35\u0026ndash;39' AND place of residence == 'rural' AND Community based health insurance == 'no' AND Then child=='Alive'\u003c/p\u003e\n\u003cp\u003eRule 2:- if currently breast feeding and preterm == 'no' AND maternal education== 'no education' AND wanted least children == 'wanted then' AND smoke ciggrate == 'no' AND health insurance provide by employer == 'no' AND smoke Tabaco == 'never in union' AND types of cooking fuel == 'wood' AND occupation == 'not working AND wealth index== 'poorest' AND maternal age == '40\u0026ndash;44' AND place of residence == 'urban' AND Community based health insurance == 'no' AND Then child=='Died'\u003c/p\u003e\n\u003cp\u003eRule3:- if currently breast feeding and preterm == 'no' AND maternal education== 'no education' AND wanted least children == 'wanted then' AND smoke ciggrate == 'no' AND health insurance provide by employer == 'no' AND smoke Tabaco == 'never in union' AND types of cooking fuel == 'wood' AND occupation == 'not working AND wealth index== 'poorest' AND maternal age == '45\u0026ndash;49' AND place of residence == 'rural' AND Community based health insurance == 'no' AND Then children=='Alive'\u003c/p\u003e\n\u003cp\u003eRule4:- if currently breast feeding and preterm == 'no' AND maternal education== 'no education' AND wanted least children == 'wanted then' AND smoke ciggrate == 'no' AND health insurance provide by employer == 'no' AND smoke Tabaco == 'never in union' AND types of cooking fuel == 'wood' AND occupation == 'not working AND wealth index== 'poorest' AND maternal age == '45\u0026ndash;49' AND place of residence == 'rural' AND Community based health insurance == 'no' AND Then children=='Alive'\u003c/p\u003e\n\u003cp\u003eRule 5:- if currently breast feeding and preterm == 'no' AND maternal education== 'no education' AND wanted least children == 'wanted then' AND smoke ciggrate == 'no' AND health insurance provide by employer == 'no' AND smoke Tabaco == 'never in union' AND types of cooking fuel == 'wood' AND occupation == 'not working AND wealth index== 'poorest' AND maternal age == '15\u0026ndash;19' AND place of residence == 'rural' AND Community based health insurance == 'no' AND Then children== 'Died'\u003c/p\u003e\n\u003c/div\u003e"},{"header":"5. Discussion","content":"\u003cp\u003eAs we discussed in risk factors identification section, the risk factors were identified using feature importance techniques and rules were generated using best performing algorithm which is gradient boosting as show in the experimental section. As we have discussed in the experimental result section, the proposed system achieved an overall performance of 90.24%, which is better a result compared to a result achieved by previous studies using gradient boosting machine learning algorithm which was 83% overall performance. We have deployed the model on cloud using Heroku and Flask framework and can be freely accessed via this link: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://perinatal-mortality.herokuapp.com/\u003c/span\u003e\u003cspan address=\"http://perinatal-mortality.herokuapp.com/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e"},{"header":"6. Conclusion","content":"\u003cp\u003eThis study aims at developing a model that predicts perinatal mortality in the case of Ethiopia by using homogeneous ensemble machine learning methods. The gradient boost algorithm has registered the highest performance with 99.72% recall, 90.24% accuracy, 92.80% f1-score, 86.96% ROC and 87.24% precision. We identified the determinant risk factors of perinatal mortality with feature importance techniques such as maternal residence, level of education, birth interval, and community-based health insurance. The most relevant rules, that helps to formulate evidence-based strategies and policies towards maintaining perinatal mortality, were generated from the best performing model, and the rules were validated by the domain experts.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eANC\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp; Antenatal Care\u003c/p\u003e\n\u003cp\u003eCBHI\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp; Community-Based Health Insurance\u003c/p\u003e\n\u003cp\u003eEDHS\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp; Ethiopia Demographic Health Survey\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;MH\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp; Maternal Health\u003c/p\u003e\n\u003cp\u003eOOP\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp; Out-Of-Pocket\u003c/p\u003e\n\u003cp\u003ePNC\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp; Postnatal Care\u003c/p\u003e\n\u003cp\u003eROC\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp; Receiver Operating Characteristics\u003c/p\u003e\n\u003cp\u003eSDGs\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp;\u0026nbsp; Sustainable Development Goals\u003c/p\u003e\n\u003cp\u003eSMOTE\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Synthetic Minority Over-Sampling Technique\u003c/p\u003e\n\u003cp\u003eWHO\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;World Health Organization\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe would like to acknowledge the Ethiopia central statistical agency for providing us the data.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe research was supported by the University of Gondar research and community service vice president's office.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets generated and/or analysed during the current study are available in the \u0026lsquo;perinatal_dataset-\u0026rsquo; repository, https://github.com/dawitemu1/perinatal_dataset-.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll methods used in this study followed guidelines and regulationsthatwere approved by the institutional reviewboardof the University of Gondar. Members of the board are Professor FelekeMoges, Mr. NiguseYigzaw, Mr. AbiyotEndale, Dr. MisayeMulate, Dr. AlemayehuTekelu and Dr. BimerewAdmasu.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors report that they have no conflicts.\u003c/p\u003e"},{"header":"References","content":"\u003cp\u003e[1] A. B. Comfort, L. A. Peterson, and L. E. Hatt, \u0026ldquo;Effect of health insurance on the use and provision of maternal health services and maternal and neonatal health outcomes: A systematic review,\u0026rdquo; J. Heal. Popul. Nutr., vol. 31, no. 4 SUPPL.2, 2013, doi: 10.3329/jhpn.v31i4.2361.\u003c/p\u003e\n\u003cp\u003e[2] V. Jain and J. M. Chatterjee, \u0026ldquo;Machine Learning with Health Care Perspective: Machine Learning and Healthcare,\u0026rdquo; no. March, 2020, doi: 10.1007/978-3-030-40850-3.\u003c/p\u003e\n\u003cp\u003e[3] R. Rasaily et al., \u0026ldquo;Effect of home-based newborn care on neonatal and infant mortality: A cluster randomised trial in India,\u0026rdquo; BMJ Glob. Heal., vol. 5, no. 9, pp. 1\u0026ndash;11, 2020, doi: 10.1136/bmjgh-2017-000680.\u003c/p\u003e\n\u003cp\u003e[4] J. R. Daw, T. N. A. Winkelman, V. K. Dalton, K. B. Kozhimannil, and L. K. Admon, \u0026ldquo;Medicaid expansion improved perinatal insurance continuity for low-income women,\u0026rdquo; Health Aff., vol. 39, no. 9, pp. 1531\u0026ndash;1539, 2020, doi: 10.1377/hlthaff.2019.01835.\u003c/p\u003e\n\u003cp\u003e[5] E. Jukes, Encyclopedia of Machine Learning and Data Mining (2nd edition), vol. 32, no. 7/8. 2018.\u003c/p\u003e\n\u003cp\u003e[6] W. Soors, N. Devadasan, V. Durairaj, and B. Criel, \u0026ldquo;Community Health Insurance and Universal Coverage: Multiple paths, many rivers to cross,\u0026rdquo; pp. 1\u0026ndash;122, 2010.\u003c/p\u003e\n\u003cp\u003e[7] N. Haven et al., \u0026ldquo;Community-based health insurance increased health care utilization and reduced mortality in children under-5, around Bwindi Community Hospital, Uganda between 2015 and 2017,\u0026rdquo; Front. Public Heal., vol. 6, no. OCT, 2018, doi: 10.3389/fpubh.2018.00281.\u003c/p\u003e\n\u003cp\u003e[8] B. J. Akombi and A. M. Renzaho, \u0026ldquo;Perinatal mortality in sub-saharan africa: A meta-analysis of demographic and health surveys,\u0026rdquo; Ann. Glob. Heal., vol. 85, no. 1, pp. 1\u0026ndash;8, 2019, doi: 10.5334/aogh.2348.\u003c/p\u003e\n\u003cp\u003e[9] P. R. Ghimire, K. E. Agho, A. M. N. Renzaho, M. K. Nisha, M. Dibley, and C. Raynes-Greenow, \u0026ldquo;Factors associated with perinatal mortality in Nepal: Evidence from Nepal demographic and health survey 2001-2016,\u0026rdquo; BMC Pregnancy Childbirth, vol. 19, no. 1, pp. 1\u0026ndash;12, 2019, doi: 10.1186/s12884-019-2234-6.\u003c/p\u003e\n\u003cp\u003e[10] Z. A. Hassan and M. J. Ahmed, \u0026ldquo;Factors associated with immunisation coverage of children aged 12- 24 months in Erbil / Iraq 2017-2018,\u0026rdquo; vol. 24, no. 08, pp. 12222\u0026ndash;12235, 2020, doi: 10.37200/IJPR/V24I8/PR281205.\u003c/p\u003e\n\u003cp\u003e[11] R. A. Knuppel and J. H. Shepherd, \u0026ldquo;Perinatal mortality rates,\u0026rdquo; Br. Med. J., vol. 280, no. 6228, p. 1376, 1980, doi: 10.1136/bmj.280.6228.1376.\u003c/p\u003e\n\u003cp\u003e[12] B. H. Jena, G. A. Biks, K. A. Gelaye, and Y. K. Gete, \u0026ldquo;Magnitude and trend of perinatal mortality and its relationship with inter-pregnancy interval in Ethiopia: A systematic review and meta-analysis,\u0026rdquo; BMC Pregnancy Childbirth, vol. 20, no. 1, pp. 1\u0026ndash;13, 2020, doi: 10.1186/s12884-020-03089-2.\u003c/p\u003e\n\u003cp\u003e[13] G. T. Debelew, \u0026ldquo;Magnitude and Determinants of Perinatal Mortality in Southwest Ethiopia,\u0026rdquo; J. Pregnancy, vol. 2020, 2020, doi: 10.1155/2020/6859157.\u003c/p\u003e\n\u003cp\u003e[14] G. Bayou and Y. Berhan, \u0026ldquo;Perinatal mortality and associated risk factors: a case control study.,\u0026rdquo; Ethiop. J. Health Sci., vol. 22, no. 3, pp. 153\u0026ndash;62, 2012.\u003c/p\u003e\n\u003cp\u003e[15] Y. Getiye and M. Fantahun, \u0026ldquo;Factors associated with perinatal mortality among public health deliveries in Addis Ababa, Ethiopia, an unmatched case control study,\u0026rdquo; BMC Pregnancy Childbirth, vol. 17, no. 1, pp. 1\u0026ndash;7, 2017, doi: 10.1186/s12884-017-1420-7.\u003c/p\u003e\n\u003cp\u003e[16] B. Demissie and K. G. Negeri, \u0026ldquo;Effect of community-based health insurance on utilization of outpatient health care services in southern ethiopia: A comparative cross-sectional study,\u0026rdquo; Risk Manag. Healthc. Policy, vol. 13, pp. 141\u0026ndash;153, 2020, doi: 10.2147/RMHP.S215836.\u003c/p\u003e\n\u003cp\u003e[17] C. Kabudula et al., \u0026ldquo;Evaluation of machine learning methods for predicting the risk of child mortality in South Africa,\u0026rdquo; 2019.\u003c/p\u003e\n\u003cp\u003e[18] G. Nguyen, \u0026ldquo;Evaluating statistical and machine learning methods to predict risk of in-hospital child mortality in Uganda,\u0026rdquo; 2016.\u003c/p\u003e\n\u003cp\u003e[19] D. D. Atnafu, H. Tilahun, and Y. M. Alemu, \u0026ldquo;Community-based health insurance and healthcare service utilisation, North-West, Ethiopia: A comparative, cross-sectional study,\u0026rdquo; BMJ Open, vol. 8, no. 8, pp. 1\u0026ndash;6, 2018, doi: 10.1136/bmjopen-2017-019613.\u003c/p\u003e\n\u003cp\u003e[20] K. Shiferaw, B. Mengiste, T. Gobena, and M. Dheresa, \u0026ldquo;The effect of antenatal care on perinatal outcomes in Ethiopia: A systematic review and meta-analysis,\u0026rdquo; PLoS One, vol. 16, no. 1 January, pp. 1\u0026ndash;19, 2021, doi: 10.1371/journal.pone.0245003.\u003c/p\u003e\n\u003cp\u003e[21] D. Prasad et al., \u0026ldquo;The effect of community-based health insurance on the utilization of modern health care services : Evidence from Burkina Faso,\u0026rdquo; vol. 90, pp. 214\u0026ndash;222, 2009, doi: 10.1016/j.healthpol.2008.09.015.\u003c/p\u003e\n\u003cp\u003e[22] I. Journal and C. Science, \u0026ldquo;Class Imbalance Problem in Data Mining : Review,\u0026rdquo; vol. 2, no. 1, 2013.\u003c/p\u003e\n\u003cp\u003e[23] R. P. Ribeiro, \u0026ldquo;SMOTE for Regression,\u0026rdquo; no. October 2015, 2013, doi: 10.1007/978-3-642-40669-0.\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-informatics-and-decision-making","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"midm","sideBox":"Learn more about [BMC Medical Informatics and Decision Making](http://bmcmedinformdecismak.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/midm/default.aspx","title":"BMC Medical Informatics and Decision Making","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"homogenous ensembles, machine learning, perinatal mortality, maternal health, health insurance. ","lastPublishedDoi":"10.21203/rs.3.rs-1445740/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-1445740/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003ePerinatal mortality in Ethiopia is the highest in Africa, with 68 per 1000 pregnancies intrapartum deaths (death during the delivery). It is mainly associated with home delivery, which contributes for more than 75% of perinatal deaths. Financial constraints have a significant impact on timely access to maternal health (MH) care. Financial incentives, such as health insurance, can address the demand- and supply-side factors. This study, hence, aims to predict perinatal mortality based on maternal health status and health insurance service using homogeneous ensemble machine learning methods\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eThe data was collected from Ethiopian demographic health survey from 2011 to 2019 G.C. The data were pre-processed to get quality data that are suitable for a homogenous ensemble machine-learning algorithm to develop a model that predicts perinatal mortality.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eFor constructing the proposed model, three experiments were conducted using random forest, gradient boosting, and cat boost algorithms. The overall accuracy of random forest, gradient boosting, and cat boost with 17 features is 89.95%, 90.24%, and 82%, respectively.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eWe finally concluded that perinatal mortality over time in Ethiopia is decreasing. We found out that perinatal mortality in Ethiopia is associated with risk factors such as community-based health insurance, mother's educational level, residence, mother age, wealth status, distance to the health facility, preterm, smoke cigarette, anemia level, haemoglobin level, and marital status.\u003c/p\u003e","manuscriptTitle":"Predicting Perinatal Mortality Based on Maternal Health Status and Health Insurance Service using Homogeneous Ensemble Machine Learning Methods","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2022-03-22 16:22:27","doi":"10.21203/rs.3.rs-1445740/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Major revision","date":"2022-06-02T15:57:46+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2022-05-31T14:27:12+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2022-03-21T12:03:49+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2022-03-21T12:01:50+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Medical Informatics and Decision Making","date":"2022-03-12T20:53:21+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-informatics-and-decision-making","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"midm","sideBox":"Learn more about [BMC Medical Informatics and Decision Making](http://bmcmedinformdecismak.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/midm/default.aspx","title":"BMC Medical Informatics and Decision Making","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"1fb6fdc7-8c18-4ac9-bc8a-3a06c5491b3c","owner":[],"postedDate":"March 22nd, 2022","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2023-10-16T18:14:36+00:00","versionOfRecord":{"articleIdentity":"rs-1445740","link":"https://doi.org/10.1186/s12911-022-02084-1","journal":{"identity":"bmc-medical-informatics-and-decision-making","isVorOnly":false,"title":"BMC Medical Informatics and Decision Making"},"publishedOn":"2022-12-28 18:07:21","publishedOnDateReadable":"December 28th, 2022"},"versionCreatedAt":"2022-03-22 16:22:27","video":"","vorDoi":"10.1186/s12911-022-02084-1","vorDoiUrl":"https://doi.org/10.1186/s12911-022-02084-1","workflowStages":[]},"version":"v1","identity":"rs-1445740","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-1445740","identity":"rs-1445740","version":["v1"]},"buildId":"oE6Zbj460LM0Up2FdVbMZ","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}