• 

  Fullpage screenshot

• 

  Input screenshot focus

• 

  Output screenshot focus

• 

  Kenkō

• 

  Kenkō App Icon

• 

  Kenkō App Icon 2

About Kenkō

Kenkō is an innovative digital health companion aimed at democratising non-critical medical expertise and alleviating pressure on overwhelmed healthcare services by providing instantaneous diagnoses and treatment suggestions. Built using PartyRock, Kenkō delivers quick, reliable, and responsible health advice for minor ailments, reducing the burden on traditional healthcare infrastructure.

Inspiration

Kenkō was inspired by the desire to bridge the gap between convenience and quality care for non-urgent medical issues.

The COVID-19 pandemic exposed weaknesses in healthcare systems globally, causing strain on frontline workers and extended waiting periods for patients. During this time, people turned to telemedicine and DIY health checks, indicating demand for accessible, responsible health advice.

What it does

Using human input and a chatbot interface for more human discussions, Kenkō receives user-reported symptoms, intelligently parses the information, cross-references verified medical databases, and returns probable diagnoses along with suggested courses of action. If the reported issue requires urgent attention, Kenkō will notify the user accordingly.

How I built it

Constructed entirely on Amazon PartyRock, I chose the Claude model to drive Kenkō's advanced natural language processing. Among the various available options, Claude demonstrated superior performance in understanding user inputs, particularly in deciphering intricate medical jargon and context clues. Its balanced emphasis on both accuracy and fluency contributed to more precise diagnoses and smoother interactions, setting the ideal foundation for a dependable digital health assistant.

Challenges I ran into

Whilst being able to navigate the vast available medical databases via Claude as a foundational model, there is responsibility required for outputs, especially given the sensitivity surrounding medical and health-related information.

Additionally I encountered difficulties in implementing multimodal input. PartyRock provides various models for text and image content generation, the lack of capability for user image upload hampered Kenkō's accessibility features. Had PartyRock accommodated multimodal input, it would have greatly benefited users struggling with text entry, opening doors to improved accessibility. Nevertheless, I adapted by focusing on enhancing Kenkō's text-based input, ensuring users still received reliable and responsible health advice for minor ailments.

Accomplishments I am proud of

Successfully implementing a sympathetic, personable chatbot interface helped establish rapport with users, while maintaining focus on accurate medical diagnosis. Moreover, connecting Kenkō to thoroughly evaluated medical databases guaranteed the provision of accurate, up-to-date health information.

What I learned

Working on Kenkō allowed me to gain firsthand experience in applying responsible AI principles to a consumer-facing product. Through countless iterations, I honed my ability to parse dense medical terminology, grasping subtle distinctions vital to forming precise diagnoses. Overall, Kenkō taught me the immense power and potential of AI in augmenting humanity's capacity for compassionate care.

What's next for Kenkō

PartyRock provided a great opportunity to MVP Kenkō but building this as a standalone app/utilising Amazon Bedrock can fortify Kenkō's position as a cornerstone of responsible, accessible healthcare. Future developments for Kenkō involve:

• Integrating into healthcare services: Integrating with medical providers, e.g. Integrating with medical history or booking Doctor appointments if required
• Multi-language capabilities: Currently Kenkō will only provide responses within English, which is a slightly limitation with the Claude model
• Multi-modal capabilities: Enabling secure image uploads could open possibilities to more responsible and accurate diagnosis and treatment.
• KenkōBot's personality: I would like to further develop the chatbot's personality. The character's honest yet patient focused was something very warming to those that have seen the movie.

Built With

• partyrock