ClearPicture

Prevention and Early Intervention track (helping people act on their health with AI tools)

Inspiration

Diagnostic errors and late intervention are often the result of incomplete information. We solve this by building a user-friendly platform that documents evidence of health complaints, while probing the user to detail relevant information, which is then relayed to clinicians.

Challenges we ran into

The focus of our project was to maintain trust in the medical system, and improve access to verifiable patient symptoms to medical professionals. As a result, we carefully designed our pipeline to avoid making a diagnosis, as AI systems -including ours, can not replace the holistic context and knowledge of a clinician from a live interaction.

Hence, the challenge was focused around planning and deploying an intelligent system that can probe for relevant information to build a log without influencing a patient's prognosis on their condition. This additionally avoids speculation and anxiety around receiving treatment, as well as providing medical professionals the facility to get structured access to detailed information, which can be used for a swift response and escalation of treatment when deemed necessary.

Accomplishments that we're proud of

Throughout, we maintained a familiar and minimalistic user interface for ease of access for all user groups. We also tested usability via Claude Desktop, Manus and the web interfaces to ensure minimal frictions for adoption.

Importantly, the main achievement is being able to contribute to enabling fruitful and reliable human-AI interaction. Our integration leverages the human expert's reasoning, as well as our system reflectively informing the clinician about the potential biases in the log. Overall, this allows for more informed decision-making by specialists to respond quicker patient conditions.

Built With

• claude
• codex
• cursor
• manus
• z.ai