Timecourse of Disease: Derived from Analysis of Unstructured Documentation

Author: Chethan Sarabu

Medicine is a chronological process. Both the daily practical function of medicine as well as the underlying diseases it is trying to treat change over time. For many diseases we actually have very limited information about the true time course of how that condition progress or regresses. The data collected in the Electronic and Patient Health Records can help elucidate these chronological pathways through an analysis of aggregated big data sets. A patient may show up to an office visit with an abnormal urine test and wants to know what is the chance they will develop kidney disease and when that may happen. The physician makes a best guest for a follow up visit but there is poor specific data on answering the patient’s question. At each visit there is chronological information charted in the note about their symptom progression and then potentially after visit documented on phone calls or patient generated data apps/portals. A time course mapping could be made of one patient’s symptom to disease and treatment progression with visits as anchor points to corroborate individually reported time data. Multiple patient profiles could be combined anchored across their diagnostic and treatment codes to then generate a true time course of disease aggregated from big patient data.