THE CHALLENGE
Our client wanted to develop an app to disrupt the healthcare space using technology, by helping junior doctors in rural areas improve their prescriptions for a chronic disease, based on disease progression. The app would be used to collect data on the symptoms of patients over multiple visits. They needed a model that can predict tweaks to the current prescription to provide relief from symptoms and arrest potential complications.
THE SOLUTION
- Our team analyzed historical prescription data from senior doctors which were deconstructed to find the underlying patterns of medication depending on the disease progression.
- These patterns were used to create an AI model
- Model change in prescription
- Automated selection of features for predicting each potential change in prescription
- Deal with sparse data for certain prescription changes using domain insights
TECHNIQUES, TECHNOLOGIES, TOOLS
- Bayesian Structure Learning, Random Forest, Gradient Boosted Decision Trees, Dynamic Bayesian Networks
RESULT/IMPACT
The app helped junior doctors in rural areas to be more consistent in their prescriptions, removing bias in the decision making. Thereby, improving the lives of the patients.
