The entire healthcare system exists to do the opposite of this 🤔

@Gigacasting Oof, you’re more jaded than I am haha

I think this is unlikely to come soon, but this would be very useful for assessing calibration of clinical judgements. It would have to be followed by a more explicit understanding, in clinicians, of how severe the consequences of false negative and false positive results would be in a given situation. Now in some contexts this is mostly done in a centralised way by a technocratic committee who determine treatment pathways and guidelines, usually including where available quantitative estimates of these figures.

@WXTJ You’re absolutely right about the false positives and negatives. I kept it simple here. But for diagnoses we never want to miss (eg, heart attack), we better be over-testing. In some such contexts, a well calibrated prediction would be one that doesn’t miss any heart attacks (which is very hard to determine). Data on how much to test is much sparser than I suspect people think. People are using big data and AI to work on this problem.

Private groups will very likely be a thing by then - theoretically, a hospital could create a private group, track predictions, and essentially do what you are suggesting.

Would private group functionality result in yes resolution, or would additional functionality / adoption be required?

@CarsonGale That’s a good question. I think something encrypted and HIPPA compliant would result in a yes. I figure this would only happen if there was indication that it would be adopted. But let’s say that no other functionality or proof of adoption is required.

@VadimShteyler @CarsonGale Private groups are one thing, but I doubt Manifold will want to get into HIPPA compliant development, it seems a bit outisde their wheelhouse, and I doubt professionals will want to reply upon Manifold for privacy compliance.