A lot of talk has been ongoing on unstructured data – as in this Facebook note, your blog post, a patient’s question on a support group, a doctor’s response on a newsgroup, a PDF scan copy of your aunt’s prescription, a journal article, its rebuttal, and you get the idea. There clearly is a lot of information pertaining to health and it may be an overkill, when it comes to decision making. Ask a cancer expert. They trust raw data, primary data from clinical trials that are high quality, randomized control. They trust their own experience treating large volumes of patients firsthand, observing the toxicities and managing the comorbidities (other health conditions like a weak heart or compromised liver) that may complicate a cancer treatment. And ultimately, data is only valuable when it is credible and lends itself to crisp, crystal clear, decision making: i.e. 1) sources of data matter (a randomized controlled clinical trial versus many doctors’ notes… or outcomes of patients at an expert cancer center versus self reported comments by patients on an online forum), and 2) structuring data in a way that doctors think from the get go, allows valuable information to be retrieved for decision making.
Hence, I am a fan of structuring data, which in and of itself is a complex challenge (Navya has developed an ontology for decision making, cancer by cancer, and that ontology is derived from the credible sources of data itself – like clinical trials and international guidelines and the handful of true experts themselves), rather than spinning cycles (computational power, manual processes, resources in time and money), on parsing mountains and mountains of any and all health related data and then trying very hard to derive accurate, usable, precise meaning from it for clinical decision making.
When it comes to cancer care, start with structured credible data and focus (spin wheels if you must) on analyzing it for decision making.