Development of ML models for a medical center

The goal of this project was to develop two machine learning models for use in a medical center. The first model is designed to predict the likelihood of patient churn, allowing the clinic to take timely measures to retain clients. The model was trained using anonymized data on patient visits, the duration of their stay at the center, the number of consultations with doctors, and other related factors

The second model was designed to automate the handling of patients' paper documentation. It was intended to scan, digitize, and process information extracted from physical documents

Both models were successfully developed and implemented in practice. The first model demonstrates a client churn prediction accuracy of 78%. The generated data is forwarded to the quality control department, where specialists analyze the reasons why clients may have discontinued services and switched to competitors.

The second model achieved an information reading accuracy of up to 99% from paper documents, enabling the medical center to effectively digitize nearly all paper documentation, including client ID cards and medical reports obtained by patients at other clinics