[Paper Review] Internet - assisted risk assessment of infectious diseases in women sexual and reproductive health

Andrzej Jarynowski, Damian Marchewka+1 more|arXiv (Cornell University)|Feb 5, 2018

Consumer Behavior and Market Dynamics3 citations

TL;DR

This paper presents open-source, internet-based risk calculators for infectious diseases in women's sexual and reproductive health, using data from hospital information systems (HIS) and patient surveys via Bayesian inference, logistic regression, and agent-based modeling. It enables real-time, personalized risk assessment for healthcare-associated infections during childbirth and sexually transmitted infections like HIV, with interactive web applications delivering dynamic risk scores and modifiable factor feedback.

ABSTRACT

We develop open source infection risk calculators for patients and healthcare professionals as apps for hospital acquired infections (during child-delivery) and sexually transmitted infections (like HIV). Advanced versions of ehealth in non-communicable diseases do not apply to epidemiology much. There is, however, no infection risk calculator in the Polish Internet so far, despite the existence of data that may be applied to create such a tool. The algorithms involve data from Information Systems (like HIS in hospitals) and surveys by applying mathematical modelling, Bayesian inference, logistic regressions, covariance analysis and social network analysis. Finally, user may fill or import data from Information System to obtain risk assessment and test different settings to learn overall risk. The most promising risk calculator is developed for Healthcare-associated infections in modes for patient hospital sanitary inspection. The most extended version for hospital epidemiologists may include many layers of hospital interactions by agent-based modeling. Simplified version of calculator is dedicated to patients that require personalized hospitalization history of pregnancy described by questions represented by quantitative and qualitative variables. Patients receive risk assessment from interactive web application with additional description about modifiable risk factors. We also provide solution for sexually transmitted infections like HIV. The results of calculations with meaningful description and percentage chances are presented in real-time to interested users. Finally, user fills the form to obtain risk assessment for given settings.

Motivation & Objective

• To address the lack of internet-based infection risk assessment tools for women's sexual and reproductive health in Poland.
• To develop accessible, data-driven risk calculators for healthcare professionals and patients using existing hospital and survey data.
• To integrate advanced statistical methods such as Bayesian inference and logistic regression into clinical decision support for infectious disease risk.
• To enable real-time, personalized risk assessment through interactive web applications with modifiable factor feedback.
• To extend the tool’s utility from patient-level risk to epidemiological modeling for hospital infection control.

Proposed method

• Utilizes data from hospital information systems (HIS) and patient surveys as input for risk modeling.
• Employs Bayesian inference to update infection risk probabilities based on observed variables.
• Applies logistic regression to estimate the probability of infection outcomes using quantitative and qualitative variables.
• Incorporates covariance analysis to assess relationships between risk factors and infection outcomes.
• Uses social network analysis to model transmission dynamics in hospital settings.
• Employs agent-based modeling in advanced versions to simulate multi-layered hospital interactions and infection spread.

Experimental results

Research questions

• RQ1 How can internet-based tools improve risk assessment for healthcare-associated infections during childbirth in a clinical setting?
• RQ2 To what extent can patient-reported data and HIS data be combined to predict infectious disease risk in reproductive health?
• RQ3 What role do modifiable risk factors play in shaping personalized infection risk profiles for pregnant patients?
• RQ4 How effective is real-time risk feedback via web applications in enhancing patient and provider decision-making?
• RQ5 Can agent-based modeling improve the accuracy of infection risk prediction in hospital environments?

Key findings

• The developed risk calculators provide real-time, personalized infection risk assessments using both HIS data and patient-reported information.
• The most promising application is for healthcare-associated infections during childbirth, with a dedicated mode for hospital sanitary inspections.
• A simplified patient-facing version enables risk assessment using a questionnaire with quantitative and qualitative variables.
• The advanced version for hospital epidemiologists supports multi-layered modeling of hospital interactions through agent-based simulation.
• Users receive dynamic risk scores with percentage chances and actionable feedback on modifiable risk factors.
• The system is open source and designed for integration into existing eHealth frameworks in non-communicable disease contexts.