Quantitative Healthcare
fcrimins edited this page Dec 16, 2016
·
15 revisions
In five years, machine learning will be a part of every doctor’s job, Vic Gundotra says
Healthcare Triage: Health Services Research, Diabetes, and the YMCA (4/28/16)
Cancer drug development time halved thanks to artificial intelligence (10/13/15)
- Berg Health’s team used a specialised form of artificial intelligence to compare samples taken from patients with the most aggressive strains of cancer, including pancreatic, bladder and brain, with those from non-cancerous individuals. The technology highlighted disparities between the corresponding biological profiles, selecting those it predicted would respond best to the drug.
Email: "Idea: ignore drugs, they're all bs"
- In that the rate of new drug discovery is massively slowing down. The real gains to come are in genetics.
- Another Alzheimer's IPO http://pipeline.corante.com/archives/2015/07/30/another_alzheimers_ipo.php
- Maybe a lot can be learned from failed drug studies if deep learning is applied because it doesn't sound like we know how drugs work. http://www.washingtonpost.com/news/wonkblog/wp/2015/07/23/one-big-myth-about-medicine-we-know-how-drugs-work/
- Or maybe research them differently: Generate random compounds and give them to mice. Then monitor the mice in tons of different ways (gene expression perhaps the most comprehensive) for effects. Then use machine learning to figure out what compounds do what. It might point someone in the right direction if nothing else.
Email: Google Healthcare
- Is Healthcare Google’s Next Big Business After Search? This Investment Bank Thinks So. http://recode.net/2015/09/13/analyst-investment-bank-cowen-report-google-healthcare-next-multi-billion-business/
Idea: EMR
- From a large set of primary care patient records, identify individuals who might have unusual health conditions. (this idea was taken from this quiz https://www.coursera.org/learn/machine-learning/exam/Dx28I/anomaly-detection)
- Perhaps we want to flag such patients in addition to the 8 maladies
Email: Do EMRs include prescribed treatment? (9/17/15)
- If so, use that to infer the disease rather than trying to predict the disease.
- Just like collaborative filtering where parameters and features are learned simultaneously.
Email: Diagnosis and prescription biases (no subject)
- Decisions, decisions, … http://theincidentaleconomist.com/wordpress/decisions-decisions/
- "How do physicians make them? To what extent are they prone to the same cognitive biases and errors that plague the rest of us? Here’s some literature on these questions."
Email: Thiel on drugs
- A Biotech Education, In Progress
- http://blogs.sciencemag.org/pipeline/archives/2015/09/14/a-biotech-education-in-progress
- "so what does he think of drug research? Not much."
- "All these companies that start as technological companies become antitechnological in character...if it turns out that these antitechnology companies are going to be good investments, that’s quite bad for our society."
- FWC - This is the same in every industry, so why not build rent seeking into the law as a line item on a financial disclosure? As soon as a company starts spending too much on rent seeking they should be broken up, just like monopolies are broken up. Both scenarios are anti-competitive.
Email: Blood Pressure, the Mystery Number
- nobody really knows what the optimal BP number is. 120, 140?
- studies don't typically control for other factors like cholesterol and exercise
- http://www.nytimes.com/2015/06/23/health/blood-pressure-the-mystery-number.html
Email: The New Old Age: Chronic Kidney Disease Can Be Dubious Diagnosis
- gist: pretty much all older adults have kidney disease, but is it a disease or are they just old?
- http://www.nytimes.com/2015/09/15/health/chronic-kidney-disease-can-be-dubious-diagnosis.html
- http://www.webmd.com/a-to-z-guides/understanding-kidney-disease-prevention