BLACK FOREST HACKATHON

🚀 VoiceMed: The Voice-Controlled Clinical Assistant

🔍 Problem Statement: In hospitals, interacting with Hospital Information Systems (HIS) is often complex, time-consuming, and requires manual data entry via keyboard and mouse. This issue leads to significant administrative overhead for medical staff, which consumes valuable time that is lost from direct patient care.

🎯 HACKATHON CHALLENGE: How might we use Artificial Intelligence, particularly language models (LLMs) and speech recognition, to radically simplify interaction with the HIS?

📊 Available Data & Resources: Participants will have access to a FHIR server with a coreset of synthetic data of patients, encounter, lab results and medication develop as an example backend of a HIS for their solution.

🏆 Success Criteria: A successful solution should demonstrate an intuitive, voice-controlled user interface and prove the technical feasibility of the architecture in a prototype.

💡 Key Considerations: Solutions should focus on standards like FHIR and MCP and (work with open source LLMs).

🚀 Synchronize music in real-time to the listener’s brain activity

🔍 Problem Statement: Unlike conventional music, neudio makes music that responds to the listener’s brain, providing gentle brain modulation to improve mental health and cognitive performance.

🎯 HACKATHON CHALLENGE: How can we implement brain-responsive music on a mobile device, i.e. iPhone or iPad? How can we use real-time brain signals streamed from a bluetooth brain-sensing device for low-latency control of existing music synthesis apps?

📊 Available Data & Resources: Participants will get access to a brain-recording device and its SDK, existing signal processing algorithms and Swift code for a macOS prototype, to develop their solution.

🏆 Success Criteria: A successful solution will be able to process brain signals and control commercial iOS music-creation software via inter-app MIDI messages.

💡 Key Considerations: Solutions should provide a low-latency (< 100 ms) between real-time brain-signals and the resultant audio.

🚀 Agent validated integrated Surgical Review: Automated Image & Video-Based Documentation

🔍 Problem Statement: In surgical environments, documentation is legally and clinically essential to track progress and ensure accountability. However, surgical staff often lack the time to manually compile detailed reports. This leads to inefficiencies and potential gaps in documentation quality and completeness.

🎯 HACKATHON CHALLENGE: How might we use AI, ML and smart agents to automatically generate surgery reports by selecting and annotating relevant images and video sequences, ensuring usability, effcient UX and efficiency for post-operative review in Operating Room vs. Remote?

📊 Available Data & Resources:

– Surgical video recordings and image datasets (including low-quality samples)
– Metadata such as timestamps and sequence numbers
– Annotation tools and APIs for image/video manipulation

🏆 Success Criteria: Automatically crop and select relevant image/video segments Rank and filter out low-quality visuals (e.g., white noise, blurry, empty frames) Enable intuitive gallery and detail views for fast selection, fast swipe selection. Support annotation and report generation Demonstrate usability on small displays (10–13” tablets) vs 24” Desktop. Maintain chronological order and allow smart relevance-based display. Test it with an Agent, by UX metrics and real User Feedback to proof products success.

💡 Key Considerations: Be highly automated and efficient Allow both on-device (post-surgery) and in-app (batch review) workflows Include a “Tinder-like” feature for surgeons to swipe/select images for reports Ensure access to all data while highlighting the most relevant content Respect safety zones in image documentation.

🚀 Vision Agent for blind Sonar: Conversational AI Assistant for Real-Time Environment Understanding

🔍 Problem Statement: Blind Sonar enables users to feel their environment haptically, but users cannot ask questions about what they’re sensing. „What’s in front of me?“, „What color is the car?“ – these questions currently remain unanswered without human assistance.

🎯 HACKATHON CHALLENGE: How might we build an edge AI-powered vision agent that blind users can talk to, receiving real-time, concise answers about their surroundings through natural conversation?

📊 Available Data & Resources:

Hardware with Camera, Mikrophone, Speaker etc.
Cloud Credits for finetuneing etc.

🏆 Success Criteria: Build a voice-controlled vision agent running on edge AI that answers questions about the environment in real-time with spatially-aware, concise responses.

💡 Key Considerations: Prioritize edge AI processing, natural language flexibility, and real-time performance while maintaining portability and battery efficiency.

🚀 Virtual Transplant Clinic / eHealth in Transplant Surgery & Care

🔍 Problem Statement: Transplant care — from pre-evaluation to surgery and long-term follow-up — is complex and requires coordinated, patient-centred management. Patients face frequent monitoring and multiple care interfaces, while immunological and transplant survival risks demand continuous assessment. Fragmented data and workflows hinder timely decisions and consistent, high-quality care.

🎯 HACKATHON CHALLENGE: How might we design digital, telemedical and AI-supported pathways and tools that improve monitoring, communication and clinical decision-making along the entire transplant journey, especially during the early post-transplant follow-up phase?

📊 Available Data & Resources: Participants will work with the CE-certified myoncare platform (clinical dashboards, patient app, digital care pathways, secure communication and device integration), and may optionally use synthetic datasets to develop additional modules or prototypes.

🏆 Success Criteria: Solutions should demonstrate a plausible path to integration into routine transplant care by improving coordination, monitoring or clinical decision-making along the transplant pathway, while remaining feasible, usable and adaptable for broader clinical implementation.

💡 Key Considerations:This challenge will be conducted on the no-code platform myoncare.

🚀 From Lifestyle to Treatment – Building a Certified Digital Care Pathway for Obesity & Type 2 Diabetes

🔍 Problem Statement: Talamed Cardio provides evidence-based lifestyle content but does not have MDR Class IIa certification. myoncare, on the other hand, is certified but does not offer motivating lifestyle content. As a result, both patients and physicians currently lack an integrated digital treatment pathway.

🎯 HACKATHON CHALLENGE: How can we combine Talamed Cardio’s lifestyle content, myoncare’s certified technology, and IoT devices (such as smartwatches, scales, and blood pressure monitors) to create an MDR Class IIa–compliant digital treatment pathway for patients with obesity and type 2 diabetes?

📊 Available Data & Resources: Access to myoncare, test accounts, and selected modules from Talamed Cardio, Anonymized datasets and sample content from both systems, IoT devices (smartwatches, scales, blood pressure monitors) including data streams, Open standards such as HL7 FHIR, as well as regulatory guidelines (MDR Class IIa, GDPR)

🏆 Success Criteria: A successful solution brings both platforms together in compliance with MDR Class IIa requirements, enables intelligent case management, delivers individually tailored programs, seamlessly integrates IoT data, and meets data protection and security requirements. The prototype should be functional within the 48-hour hackathon and leave room for creative extensions.

💡 Key Considerations:This challenge will be conducted on the no-code platform myoncare.