23 ms·

The article doesn't mention it, but I'm curious if Watson involved in this (or similar efforts) at all.

by kornork 12y ago

bayesianhorse 12y ago
I sure hope not. Associating genetic variations with symptoms or physiological findings is a relatively clear-cut statistical problem. The major challenge in whole genome association studies (or their extension to whole genome sequencing data) is to get enough data to draw meaningful conclusions. There is a cognitive dissonance in biology/medicine: On the one hand they complain about too much data to process, but at publication time, the power is too low...
- pinkyand 12y ago
  The problem with a lot of the data in electronic medical records is that they are not structured but in natural language. This is exactly where Watson can help.
  - bayesianhorse 12y ago
    I don't think so. It is too expensive to enroll more than a couple of thousand of participants for a GWA study, though they are often pooled for a broader analysis. Anyway, genotyping will probably cost at least $100 (depending on the technology a lot more), which puts manual entry of patient data into perspective. An unpaid intern can be cheaper than fancy artificial intelligence!
gwern 12y ago
No; the phenotype data was extracted from structured interviews: "Ninety-three clinical features of schizophrenia from interviews based on the Diagnostic Interview for Genetic Studies (30), as well as the Best Estimate Diagnosis Code Sheet submitted GAIN/non-GAIN to dbGaP, were initially considered with the MGS sample (see references 31, 32; see also Appendix I in the online data supplement). The Diagnostic Interview for Genetic Studies was utilized for the Portuguese Island samples. Corresponding features were extracted in CATIE from the Positive and Negative Syndrome Scale, the Quality of Life Questionnaire, and the Structured Clinical Interview for DSM-IV (23). These phenotypic sets and their relations with one another characterize the phenotypic architecture of schizophrenia (Figure 1B)." They did consider but reject using medical records (which is where a Watson might come into play): "Despite the detailed phenotypic information we had available about subjects, there are still limitations to data obtained even from reliable structured interviews like the Diagnostic Interview for Genetic Studies. Interview data are based on self-reports that are interpreted and coded by interviewers. Subjects may not be willing or able to report their symptoms accurately. We had obtained information from treatment records and family history reports, but we chose not to use such additional information, except for the resulting best-estimate final DSM ratings of diagnosis, because its extent and quality varied in unmeasured ways between cases. The greatest limitation in the phenotypic assessments in available GWAS databases has been the overreliance on subjective symptoms with an absence of objective measurements, such as cognitive tests, brain electrophysiology, and neuroimaging (37). Subjective symptoms are fuzzy indicators of the underlying pathophysiology. Objective measures could complement the assessment of symptoms and could be applied to both cases and controls, thereby providing a more comprehensive and valid characterization of the phenotype of all subjects. The biggest challenge in GWAS is access to studies with rich phenotypic data about both subjective and objective measures obtained systematically from all subjects."

The article doesn't mention it, but I'm curious if Watson involved in this (or similar efforts) at all.

by kornork 12y ago

The article doesn't mention it, but I'm curious if Watson involved in this (or similar efforts) at all.

I sure hope not. Associating genetic variations with symptoms or physiological findings is a relatively clear-cut statistical problem. The major challenge in whole genome association studies (or their extension to whole genome sequencing data) is to get enough data to draw meaningful conclusions. There is a cognitive dissonance in biology/medicine: On the one hand they complain about too much data to process, but at publication time, the power is too low...

pinkyand 12y ago
The problem with a lot of the data in electronic medical records is that they are not structured but in natural language. This is exactly where Watson can help.
- bayesianhorse 12y ago
  I don't think so. It is too expensive to enroll more than a couple of thousand of participants for a GWA study, though they are often pooled for a broader analysis. Anyway, genotyping will probably cost at least $100 (depending on the technology a lot more), which puts manual entry of patient data into perspective. An unpaid intern can be cheaper than fancy artificial intelligence!

gwern 12y ago

No; the phenotype data was extracted from structured interviews: "Ninety-three clinical features of schizophrenia from interviews based on the Diagnostic Interview for Genetic Studies (30), as well as the Best Estimate Diagnosis Code Sheet submitted GAIN/non-GAIN to dbGaP, were initially considered with the MGS sample (see references 31, 32; see also Appendix I in the online data supplement). The Diagnostic Interview for Genetic Studies was utilized for the Portuguese Island samples. Corresponding features were extracted in CATIE from the Positive and Negative Syndrome Scale, the Quality of Life Questionnaire, and the Structured Clinical Interview for DSM-IV (23). These phenotypic sets and their relations with one another characterize the phenotypic architecture of schizophrenia (Figure 1B)." They did consider but reject using medical records (which is where a Watson might come into play): "Despite the detailed phenotypic information we had available about subjects, there are still limitations to data obtained even from reliable structured interviews like the Diagnostic Interview for Genetic Studies. Interview data are based on self-reports that are interpreted and coded by interviewers. Subjects may not be willing or able to report their symptoms accurately. We had obtained information from treatment records and family history reports, but we chose not to use such additional information, except for the resulting best-estimate final DSM ratings of diagnosis, because its extent and quality varied in unmeasured ways between cases. The greatest limitation in the phenotypic assessments in available GWAS databases has been the overreliance on subjective symptoms with an absence of objective measurements, such as cognitive tests, brain electrophysiology, and neuroimaging (37). Subjective symptoms are fuzzy indicators of the underlying pathophysiology. Objective measures could complement the assessment of symptoms and could be applied to both cases and controls, thereby providing a more comprehensive and valid characterization of the phenotype of all subjects. The biggest challenge in GWAS is access to studies with rich phenotypic data about both subjective and objective measures obtained systematically from all subjects."