Yesterday, I was surprised when I treated a patient (for a heart attack) to find that he later ended up in the ER. I’ve only treated a handful (exactly five!) with MI-like symptoms, and none of them had his low heart rate or controlled breathing. Surprise meant my model needed to be updated, and a quick look at this showed me that I really only needed two symptoms to be wary.
Physicians and other healthcare providers in the ‘algorithmic’camp (vs ‘clinical’) essentially forecast diagnosis through probability-based predictions. Algorithms are also referred to simply as ‘flowcharts’ and as part of standard operating procedure. Medical algorithms incorporate a large number of established heuristics in a standardized manner, and have been shown to dramatically increase diagnostic accuracy. The medical branch of the military has access to a wealth of patient data, and they create algorithms based on that data.
In the private sector, there’s Medal, Apervita, Syapse and others. There’s a few private companies ‘democratizing healthcare data’ for a price. There’s even more information from health insurance providers, which tend to have their own healthcare data companies, which organizations can also access for a price.
DXplain uses Bayesian logic for diagnostics, and is open to physicians. TXdent does the same for dental care. Adjuvant! is publicly available to healthcare providers and exists for cancer patients. eMedicine is pretty great, but nigh useless to the layperson, other than a better version of WebMD, even though it’s a service offered by the same company. If you wanted to improve self-care, you might be able to get some mileage out of CATmaker, but it assumes you’re a provider, and I doubt a layperson would get use out of it. Tripdatabase is a curated database, mostly with links to studies from NIH, so you’ll encounter the paywall either way.
There’s also CDSS, which, while acknowledged as effective, is having problems with implementation due to the state of IT in healthcare.
Disease models exist primarily to forecast infection rates and risks. In addition to census data, there are publicly available datasets for infectious diseases.
All this to say that I suspect the specialists in your post are obfuscating the problem.
Medical datasets are not freely available to every medical institution.
Not every provider employs information gleaned from this data.
Algorithmic, probability based care is essentially controlled by a few companies.
Solve the problem, and you close part of the gap in English speaking countries. If you can read Dutch or Russian, you might be able to get access to all of the above (albeit with more geographically limited datasets) for free, but I don’t really know.
If you want to identify health risks for yourself, cross reference the probability of infection in your meatspace community with your own demographic (parents, habits, location, age, general health, medical history) information and take appropriate preventive measures.
Yesterday, I was surprised when I treated a patient (for a heart attack) to find that he later ended up in the ER. I’ve only treated a handful (exactly five!) with MI-like symptoms, and none of them had his low heart rate or controlled breathing. Surprise meant my model needed to be updated, and a quick look at this showed me that I really only needed two symptoms to be wary.
Physicians and other healthcare providers in the ‘algorithmic’ camp (vs ‘clinical’) essentially forecast diagnosis through probability-based predictions. Algorithms are also referred to simply as ‘flowcharts’ and as part of standard operating procedure. Medical algorithms incorporate a large number of established heuristics in a standardized manner, and have been shown to dramatically increase diagnostic accuracy. The medical branch of the military has access to a wealth of patient data, and they create algorithms based on that data.
In the private sector, there’s Medal, Apervita, Syapse and others. There’s a few private companies ‘democratizing healthcare data’ for a price. There’s even more information from health insurance providers, which tend to have their own healthcare data companies, which organizations can also access for a price.
DXplain uses Bayesian logic for diagnostics, and is open to physicians. TXdent does the same for dental care. Adjuvant! is publicly available to healthcare providers and exists for cancer patients. eMedicine is pretty great, but nigh useless to the layperson, other than a better version of WebMD, even though it’s a service offered by the same company. If you wanted to improve self-care, you might be able to get some mileage out of CATmaker, but it assumes you’re a provider, and I doubt a layperson would get use out of it. Tripdatabase is a curated database, mostly with links to studies from NIH, so you’ll encounter the paywall either way.
There’s also CDSS, which, while acknowledged as effective, is having problems with implementation due to the state of IT in healthcare.
Disease models exist primarily to forecast infection rates and risks. In addition to census data, there are publicly available datasets for infectious diseases.
All this to say that I suspect the specialists in your post are obfuscating the problem.
Medical datasets are not freely available to every medical institution.
Not every provider employs information gleaned from this data.
Algorithmic, probability based care is essentially controlled by a few companies.
Solve the problem, and you close part of the gap in English speaking countries. If you can read Dutch or Russian, you might be able to get access to all of the above (albeit with more geographically limited datasets) for free, but I don’t really know.
If you want to identify health risks for yourself, cross reference the probability of infection in your meatspace community with your own demographic (parents, habits, location, age, general health, medical history) information and take appropriate preventive measures.