The list of topical topics of thesis topics — Bachelor. Topics are grouped by supervisors. The student can choose the proposed topic or agree his own with the supervisor.
Bychkov Oleksii Serhiiovych
Cross-platform mobile client of the Dr.Case system with minimal functionality: quick diagnosis, review of the symptom base, session history. Localization of Ukr+Eng.
Implementation of the bot on the python-telegram-bot. The user describes the symptoms in free text → bot extracts the symptoms through /api/symptoms/extract → gives the top 5 hypotheses → offers clarification. Disclaimer about non-clinical prescription.
Development of a spa application: entering symptoms through autocomplete, obtaining hypotheses from the Dr.Case rest API, displaying progress bars of probabilities, interactive questions (Yes/No/Don't know). React + TanStack Query + Tailwind CSS.
Display usage statistics: distribution of diagnoses, average session length, success rate. Data from the Dr.Case rest API. Chart.js or Recharts.
Generation of a PDF report with the results of the diagnostic session: symptoms, question history, final rating of hypotheses with graphs, disclaimer. Using WeasyPrint or ReportLab.
Pygame 2D simulator with 10-20 agents managed by behavior trees (Sequence/Selector/Parallel). Scenarios: patrol, search, follow leader. Visualization of the state of the tree.
Classification: normal vs pneumonia on chest X-ray. Transfer learning from ResNet50 on the ChestX-ray14 dataset. Let's go to Streamlit and upload the snapshot.
Integration of Whisper API/OpenAI for voice transcription in Ukrainian. Parsing decrypted text through the Dr.Case NLP module. Demo scenario of the full session.
Interactive 15×15 grid som color-coded by disease category (infectious/cardiac/neurological). When pointing at a unit — a list of 5+ diseases. D3.js or React+Canvas.
Binary classifier on gradient boosting (XGBoost/LightGBM) for CBC parameters. Experimental accuracy assessment, ROC curve. Let's go web to Streamlit.
2D simulator in Pygame:agents- "doctors" with different specializations, exchange "hypotheses" (rules), agree on solutions. Inspiration is a microcosm of Dr.Case multi-agent diagnostics.
Basic implementation of Raft in Python with messages via ZeroMQ or gRPC. Experimental study of behavior in case of loss of messages and node failures.