[lowglow]

Give participants (patients, doctors) access to subscribe to health data or symptoms (input) and create a consensus of possible treatment plans, iterate through those plans ordered by some metric (reputation, time, cost, availability).

When cure confirmed by patient, diagnosis is proven correct, data is captured, reputation increased among participants in consensus that were correct.

You could even use this as human-in-the-loop machine learning model training for open models.

If you want to explore what it might take to build this, we've just launched an alpha of our data/ml network that can facilitate a dapp built on top of Synapse https://synapse.ai/

Happy to reach out and chat more.

[nradov]

This is ridiculous and completely unrealistic. Most conditions are never really cured and diagnoses are seldom "proven correct". We only have varying levels of improvement and confidence. Patients generally lack the skills to confirm anything themselves, especially because most of them don't understand causality. Did my condition improve because of my physician's treatment, or in spite of it?

[lowglow]

How would you improve the pipeline?

[nradov]

Increase funding for large scale clinical studies so that we have sufficient data to develop evidence-based medicine guidelines for a wider range of conditions.

[neuro_imager]

> Give participants (patients, doctors) access to subscribe to health data or symptoms (input) and create a consensus of possible treatment plans, iterate through those plans ordered by some metric (reputation, time, cost, availability).

I'm still not sure I understand what you mean. Could you give an example?

> When cure confirmed by patient, diagnosis is proven correct, data is captured, reputation increased among participants in consensus that were correct.

It's not that simple. Most patients are not either 'cured' or 'not cured'. There are multiple possible outcomes at each stage of diagnosis and management.

> You could even use this as human-in-the-loop machine learning model training for open models.

Doubtful. Current machine learning algorithms are painfully inept when given clinical data (outside of very limited use cases). I'd love to be proven wrong though.

> If you want to explore what it might take to build this, we've just launched an alpha of our data/ml network that can facilitate a dapp built on top of Synapse https://synapse.ai/ Happy to reach out and chat more.

Sure, email?

[lowglow]

dan [at] synapse [dot] ai