Global Incidence of Diarrheal Diseases-An Update Using an Interpretable Predictive Model Based on XGBoost and SHAP: A Systematic Analysis.

Background: Diarrheal disease remains a significant public health issue, particularly affecting young children and older adults. Despite efforts to control and prevent these diseases, their incidence continues to be a global concern. Understanding the trends in diarrhea incidence and the factors influencing these trends is crucial for developing effective public health strategies.

Objective: This study aimed to explore the temporal trends in diarrhea incidence and associated factors from 1990 to 2019 and to project the incidence for the period 2020-2040 at global, regional, and national levels. We aimed to identify key factors influencing these trends to inform future prevention and control strategies.

Methods: The eXtreme Gradient Boosting (XGBoost) model was used to predict the incidence from 2020 to 2040 based on demographic, meteorological, water sanitation, and sanitation and hygiene indicators. SHapley Additive exPlanations (SHAP) value was performed to explain the impact of variables in the model on the incidence. Estimated annual percentage change (EAPC) was calculated to assess the temporal trends of age-standardized incidence rates (ASIRs) from 1990 to 2019 and from 2020 to 2040.

Results: Globally, both incident cases and ASIRs of diarrhea increased between 2010 and 2019. The incident cases are expected to rise from 2020 to 2040, while the ASIRs and incidence rates are predicted to slightly decrease. During the observed (1990-2019) and predicted (2020-2040) periods, adults aged 60 years and above exhibited an upward trend in incidence rate as age increased, while children aged < 5 years consistently had the highest incident cases. The SHAP framework was applied to explain the model predictions. We identified several risk factors associated with an increased incidence of diarrhea, including age over 60 years, yearly precipitation exceeding 3000 mm, temperature above 20 °C for both maximum and minimum values, and vapor pressure deficit over 1500 Pa. A decreased incidence rate was associated with relative humidity over 60%, wind speed over 4 m/s, and populations with above 80% using safely managed drinking water services and over 40% using safely managed sanitation services.

Conclusions: Diarrheal diseases are still serious public health concerns, with predicted increases in the incident cases despite decreasing ASIRs globally. Children aged < 5 years remain highly susceptible to diarrheal diseases, yet the incidence rate in the older adults aged 60 plus years still warrants additional attention. Additionally, more targeted efforts to improve access to safe drinking water and sanitation services are crucial for reducing the incidence of diarrheal diseases globally.