Machine Learning for Healthcare and Life Sciences

What-if scenario analysis for policy makers

What-if scenario analysis

Together with a large global team of experts in IBM, we have embarked on a journey to help monitor, treat, and prevent cervical cancer in Africa. Our goal is to improve monitoring and decision-making using an integrated solution that addresses policy makers and caregivers. The idea is to promote a proactive approach to public health.

We developed a tool that blends cloud and mobile technologies with advanced analytics to gather, manage, analyze, and visualize healthcare population data for policy decision support. We showcased the technology on different use cases, including cervical cancer prevention in Kenya, and monitoring and spread prevention of vector borne diseases in Africa, such as lymphatic filariasis.

For the analytics component of the tool we developed a Dynamic Bayesian Network and used state-of-the-art machine learning algorithms to infer the correct association between demographic or environmental variables with disease factors such as progression or geographical spread.

The model is then used to simulate complex 'what-if' scenarios to determine the best outcome of where to invest efforts and resources to further support population health policy planning. The tool allows users to explore simulations of various policies, to view and compare interventions spanning multiple variables, timepoints, and locations. The design’s modular architecture and data representation separates the modelling methods, the outcome measures calculations, and the visualizations, making each component easily replaceable. These advantages make it extremely versatile and suitable for multiple uses.

For more info on the cervical cancer use case see: movie, brochure, press release.

Michal Rosen-Zvi, Lavi Shpigelman, Alan Kalton, Omer Weissbrod, Saheed Akindeinde, Soren Benefeldt, Andrew Bentley, Terry Everett, Joseph Jajinskiji, Emmanuel Kweyu, Chalapathy Neti, Joe Saab, Osamuyimen Stewart, Malcolm Ward, Guo Tong Xie, Estimating the Impact of Prevention Action: A simulation Model of Cervical Cancer Progression
Studies in health technology and informatics 2014; 205, 288-292

For more info on the lymphatic filariasis use case see:

Michal Chorev, Lavi Shpigelman, Peter Bak, Avi Yaeli, Edwin Michael, Ya’ara Goldschmidt. Advising on Policy Using a Data-Driven Policy Decision Support Tool. MedInfo2017
Lavi Shpigelman , Michal Chorev, Zeev Waks, Ya’ara Goldschmidt, Edwin Michael. Epidemiological models without process noise are probably over confident. Informatics for Health 2017