Healthcare Predictive Analytics Tool
Challenge
In something as complex and people-centric as healthcare, sometimes every minute counts. Each late decision is a missed opportunity to spot signs of deterioration and take immediate action to save a patient's life.
Therefore, physicians always need to stay one step ahead of the situation to detect and prevent critical conditions as soon as possible. It would help reduce patient risk, greatly improve outcomes, and lower overall costs.
Solution
The system uses machine learning algorithms trained on subsets of historical data collected from CMS 1500 claims to identify patients who may require:
- surgery within the next 1, 3, and 6 months,
- hospitalization, or
- a certain medication.
Project Results
Softarex’s Predictive Analytics Tool has brought healthcare providers the following results:
- 38% accuracy of surgery prediction;
- 98% NPV value;
- Enhanced patient outcomes;
- Reduced readmissions;
- Enhanced staff planning;
- Optimized day-to-day hospital operations.
Technical challenges
- Development of complex algorithms for data processing in text format;
- Development of algorithms and methods for processing thousands of patient records;
- Development of data mining algorithms to extract valuable data from patients' medical records.