Changes

==== User could arrive at wrong conclusions ====

==== User could arrive at wrong conclusions ====

Wrong conclusions could be dangerous to user's health. User should of course double check with experts like doctors and veterans of the community (and this wiki). Learning enough about health and statistics takes time that many do not have.  

Wrong conclusions could be dangerous to user's health. User should of course double check with experts like doctors and veterans of the community (and this wiki). Learning enough about health, experimentation,<ref>https://www.frontiersin.org/articles/10.3389/fdgth.2020.00003/full</ref> and statistics takes time that many do not have.  

Analysis algorithms are either hard to use or too incapable. For example, article in Nature where ML was used in multiple N-of-1 studies and that approach is both incomplete and difficult for the average user.<ref>https://www.nature.com/articles/s41398-021-01445-0</ref> [[Finding relations between variables in time series|All data aggregators]] for self tracking, besides OH, use linear regression or nothing at all. This problem can sometimes be avoided with [[Experiment VS Observational study|careful experimental design]] like RCT.  

Analysis algorithms are either hard to use or too incapable. For example, article in Nature where ML was used in multiple N-of-1 studies and that approach is both incomplete and difficult for the average user.<ref>https://www.nature.com/articles/s41398-021-01445-0</ref> [[Finding relations between variables in time series|All data aggregators]] for self tracking, besides OH, use linear regression or nothing at all. This problem can sometimes be avoided with [[Experiment VS Observational study|careful experimental design]] like RCT.