- Customer: Medical supporting staff
- Goal: Organize medical transcriptions with right speciality category
- Pains:: Manual organizing transcription is a hectic and time consuming process
- Gains:: Speed up the process of organizing and can also help in searching medical transcriptions
- Product: The primary product is to categorize transscriptions into different medical speciality, the second version could be building a search mechanism to search customer medical transcription faster.
- Alleviates: Automating categoization of transcriptions will reduce the efforts of medical staff for any discovery purpose
- Advantages: There will be no manual reading process necessary to choose the right medical speciality for the medical staffs.
- Create an automated pipeline to organize medical transcription into it's correct speciality
- Achieve > 85% for the classification problem
- Build a search mechanism as a downstream feature on top of the classification to search medical transcriptions based on keywords
- Core Features
- ML service to classify the incoming medical transcription
- Feedback loop from the medical staff to identify incorrect classifications
- Filter incorrect classifications with high confident score
- Secondary Features
- Build a search engine UI, to search transcriptions on demand
- Integrate speciality classification ML service with search engine to reduce the search space
- Constraints
- Maintain low latency
- Identify duplication of medical transcriptions