6 ms·

Is it really too "hard" to perform rigorous statical analysis? Why can't you factor in patient genetics + background + circumstances to come up with some expect

• rscho 5y ago
  The number of factors you'd have to consider to achieve that is huge enough that it makes it completely unrealistic, both practically and financially speaking. In fact, I'm 100% convinced that it's currently impossible to build such a system, given the extent of medical knowledge.
  • grog454 5y ago
    I'm skeptical. I concede it's probably a hard problem but there is an entire field dedicated to hard statistical problems called data science. What is the point of having detailed patient histories and data if it can't be used to inform decisions?
    • rscho 5y ago
      If data science was that efficient on healthcare records, we'd know it by now and everyone would do it. There is no 'big healthcare data'. We gather mostly noise, and records are full of blatant mistakes. Medicine is still more art than science today, and records mostly give you a 'feeling' for the patient's condition in light of your MD education, with a little hard data sprinkled on top.
      • stuckonempty 5y ago
        This is a fallacy. At any point in history you can say “if X field was so good, we’d have Y by now”. In 1925 you could’ve said, “if biology’s understanding of bacteria is so good, we’d have antibiotics by now”. Within 5 years, they did. There is certainly noise in healthcare data especially when patient-reported, but is it noise to say that a patient having X procedure later does or doesn’t have serious complications? Analyzes of medical care and their consequences can be evaluated and it’s not noise And big healthcare data has lagged, partially because privacy concerns trump sharing. There are companies selling anonymized medical records for basically every American now though. Big data is coming
        • rscho 5y ago
          > Big data is coming Big _bad_ data... Let's see how we fare in 5y, then. My prediction as a clinician with a special interest in stats: close to zero medical progress. But insurance priced by a ML algorithm, and much greater efficiency in coverage and claim denials.
          • nradov 5y ago
            Due to the Affordable Care Act (Obamacare), medical insurers have very little flexibility in pricing policies. There's not much point in using ML for pricing. https://www.healthcare.gov/how-plans-set-your-premiums/ https://www.healthcare.gov/how-plans-set-your-premiums/ The more likely use case for ML is detecting insurance fraud patterns.
            • webmaven 5y ago
              > The more likely use case for ML is detecting insurance fraud patterns. I think that's covered under "much greater efficiency in coverage and claim denials."
        • oaktrout 5y ago
          The first flu vaccine came about in 1945. Knowing as much as we did about viruses then, you might think we would have a cure for influenza (or the common cold) by now. Here we are almost 80 years later... big data may be coming but if takes that long it won't be in my lifetime.
      • BurningFrog 5y ago
        My impression is that data science on healthcare records is mostly illegal due to HIPAA and other regulations.
        • sxg 5y ago
          It's not illegal. It's just very complicated to do correctly, and you risk large fines if you do it incorrectly.
          • BurningFrog 5y ago
            For most practical purposes, that's has the same effects.
    • nradov 5y ago
      We mostly don't have detailed patient histories. Most of the relevant clinical data that would be needed for such a rigorous statistical analysis isn't recorded as discrete data elements, it's just unstructured text. NLP can be used to extract concept codes but the error rate is high so you still need an experienced (expensive) human to manually fix the errors. No one wants to pay for that. The codes we do have are mostly CPT4 and ICD-10 for billing purposes. Those are generally pretty accurate, but not detailed enough to reliably assess whether one surgeon is better than another at a particular procedure.
• sxg 5y ago
  Yes, it's absolutely very difficult to do rigorous statistical analysis. Genetics is oversimplified to non-physicians. It's cool that we can diagnose and predict the likelihood of getting Huntington's disease using our knowledge of genetics, but extremely few diseases are this simple. There are huge swaths of the human genome that we don't understand but are likely playing some important role in the regulation of other genes and diseases. We are nowhere close to being able to look at a patient's genome to predict anything useful outside of a handful of exceptions. Patient histories are honestly often garbage—I say that as a physician. I look through dozens of patients' charts every day, and there are constantly errors, incomplete documentation, and fragmented records across multiple institutions. Just last week I read a chart for a patient who had a documented hysterectomy from years ago. The brand new CT scan I saw showed a perfectly normal uterus. Once something goes in a patient's history it's nearly impossible to correct or remove. If some doctor from ages ago said the patient is allergic to medication X, but the patient denies it, what do I do? Usually, we opt to leave the allergy listed out of fear of the consequences if the patient is wrong.
  • grog454 5y ago
    Are you aware of any attempts to use machine learning in medical analysis and outcome prediction? I feel like this is one of the few applications where it could shine. I have no formal training in data science, but everything I've read so far seems to indicate that noisy and unreliable data is not an insurmountable problem. All of this talk about how "hard" the statistical analysis is, is strange to me. Maybe "advanced" would be a better term? If you get a patient with a contradictory medical history that somehow also contradicts what they are telling you, simply adjust your expected chance of success appropriately (to zero perhaps). In that extreme case, if you get a good outcome, congrats you got lucky. If you don't, it should have 0 impact on how you are evaluated as a doctor.
    • rscho 5y ago
      Statistics are not the problem. The data is. 'Big healthcare data' does not exist. Building what you are thinking of would require huge data gathering capabilities, that are very clearly out of reach. Most big medical co do a lot of data science (Kaiser and others). Very efficient from a managerial pov. Totally useless, medically speaking.
• Fomite 5y ago
  As someone who works on the methodological basis for a number of "Observed/Expected" healthcare metrics in the infection control and antibiotic stewardship space - just, this is pretty hard, even if you are trying to do a rigorous statistical analysis.
• lazide 5y ago
  Yup, most of the data is junk, and even if the data isn’t junk - it usually doesn’t at all mean what you think it means.