Modelling seizure-related predictors of epilepsy diagnostic gap in two urban informal settlements of Nairobi using machine learning

Aug 5, 2026

Upload by Ophelia

Epilepsy diagnosis shows a substantial diagnostic gap, particularly in low- and middle-income countries where specialist services and diagnostic capacity are limited. Using machine learning models, the study identifies seizure-related factors linked to the epilepsy diagnostic gap within the Nairobi Urban Health and Demographic Surveillance System (NUHDSS), Kenya, to guide practical community-level interventions. Data come from a two-stage census with questionnaire screening and neurologist-confirmed assessment. Latent class analysis and seven trained models evaluate symptom patterns and diagnostic risk, with gradient boost and random forest achieving high performance (AUC 98%).

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages9

Reading Time23 min

AccessFree to download

Background

Methods

Data source and study design

Latent class analysis and machine learning models

Results

Diagnostic gap by epilepsy type

Factors associated with wider diagnostic gap

Conclusion

Implications for reducing the diagnostic gap

Global Epidemiology 11 (2026) 100241 Contents lists available at ScienceDirect Global Epidemiology journal [homepage:](homepage: www.sciencedirect.com/journal/global-epidemiology)[ www.sciencedirect.com/journal/global-epidemiology](homepage: www.sciencedirect.com/journal/global-epidemiology) Modelling seizure-related predictors of epilepsy diagnostic gap in two urban informal settlements of Nairobi using machine learning Daniel Mwangaa,b,c,*, Frederick Murunga Wekesaha,d, Frank Oumaa, Symon M. Kariukia,c,e, Joan Kinuthiaa, Peter Otienoa,f,g, Thomas Kwasah, Quincy Mongare h, Abigael Machukah, Steve Cygua, Samuel Iddia, Gabriel Davis Jones c,i, Arjune Senc,j, Charles R. Newton c,e,k, Gershim Asikia,l, Damazo T. Kadengyea, for the EPInA Study Group1 a Research Division, African Population and Health Research Center, Nairobi, Kenya b Department of Mathematics, University of Nairobi, Nairobi, Kenya c Center for Global Epilepsy, University of Oxford, Oxford, United Kingdom d Julius Global Health, Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht University, the Netherlands e Neurosciences Department, Kenya Medical Research Institute Wellcome Trust Research Programme, Kilifi, Kenya f Department of Global Health and Population, Lown Scholars Program, Harvard T.H. Chan School of Public Health, Boston, MA, United States of America g Department of Public & Occupational Health, Amsterdam UMC Locatie AMC, Amsterdam, the Netherlands h Department of Clinical Medicine and Therapeutics, University of Nairobi, Nairobi, Kenya i Oxford Digital Health Labs, Nuffield Department of Women's and Reproductive Health, University of Oxford, Oxford, United Kingdom j Department of Clinical Neurosciences, University of Oxford, Oxford, United Kingdom k Department of Psychiatry, University of Oxford, Oxford, United Kingdom l Department of Women's and Children's Health, Karolinska Institute, Stockholm, Sweden A R T I C L E I N F O Keywords: Convulsive epilepsy Non-convulsive epilepsy Epilepsy diagnostic gap Machine learning Nairobi urban health and demographic surveillance system (NUHDSS) Seizures Treatment gap A B S T R A C T Background: There is a wide gap in epilepsy diagnosis, particularly in low-and middle-income countries. We used machine learning models to identify seizure-related factors associated with the epilepsy diagnostic gap within the Nairobi Urban Health and Demographic Surveillance System (NUHDSS), Kenya, to inform effective communitylevel interventions. Methods: Data were drawn from a two-stage, population-based census. In Stage-I, 56,425 residents of NUHDSS were screened for possible convulsive and non-convulsive epilepsy using a standardized questionnaire. In StageII, individuals who screened positive were invited for clinical assessment and diagnostic confirmation by neurologists. We used latent class analysis to classify symptom patterns. Seven machine learning models were trained, with extreme gradient boost and random forest models achieving the highest area under the receiver operating characteristic curve (98 %). Results: A total of 528 individuals were diagnosed with epilepsy, among whom 80 %(n = 420) had not been previously diagnosed. The epilepsy diagnostic gap was 100 %(n = 160/160) in persons with non-convulsive epilepsy, meaning that none of them had been diagnosed before the survey. Among those with convulsive epilepsy, the diagnostic gap was 71 %(n = 260/368). Experiencing fewer types of seizure symptoms, nonconvulsive seizures, or seizures with subtle features, such as those involving only one body part and those whose first experience of a seizure was recent, were associated with a wider epilepsy diagnostic gap. Conclusion: There is critically huge diagnostic gap for epilepsy in Nairobi's informal settlements. People with subtle, fewer or less obvious seizure types are more likely to be undiagnosed. These findings highlight the importance

FAQs

What problem does the study address regarding epilepsy care?

The study targets the wide epilepsy diagnostic gap, especially in low- and middle-income settings with limited diagnostic and care services.

How were seizure-related factors analyzed in this research?

Stage-I used standardized screening for possible convulsive and non-convulsive epilepsy, and Stage-II used neurologist clinical assessment and diagnostic confirmation. Latent class analysis was used to classify symptom patterns, and seven machine learning models were trained to identify predictors.

Which seizure features were linked to a wider diagnostic gap?

Fewer types of seizure symptoms, non-convulsive seizures, and subtle seizure features—such as involving only one body part or having a recent first seizure experience—were associated with a wider diagnostic gap.

Aug 5, 2026

Upload by Ophelia