30 d’abril 2018

Medicine as a data science (2)

The Evolution of Patient Diagnosis: From Art to Digital Data-Driven Science

Currently medical diagnosis is driven by a standard way to proceed. We could say that the pattern of the decision flow has not changed for years.
A physician takes a history, performs an examination, and matches each patient to the traditional taxonomy of medical conditions. Symptoms, signs, family history, and laboratory reports are interpreted in light of clinical experience and scholarly interpretation of the medical literature.
Data availability, and specifically genetic data could change completely diagnostic process.
Initiatives to develop genetic reference data at the population level could be grouped into 3 categories.First are well-known databases of genotype-phenotype relationships
as observed and submitted by researchers (eg, Online Mendelian Inheritance in Man, ClinVar, and the National Human Genome Research Institute’s Genome-Wide Association Study [GWAS] Catalog). Second are databases, such as the Genome Aggregation Database (gnomAD), the next iteration of the ExomeAggregation Consortium (ExAC) database, and the 1000 Genomes Project, that aggregate sequences
collected from other studies for secondary use. Third, patients and other study participants are invited to donate data to registries like GenomeConnect or enroll
in cohorts like the National Institutes of Health All of Us initiative, which is recruiting 1 million patients to contribute biological samples and EHR data for research.
The reference to these databases is crucial to understand what's going on in US medicine, and how european medicine stands behind.
JAMA article develops the concept of Clinical Information Commons:
There should be a new compact between patients and the health system, such that captured data and biospecimen by- products of the care deliverysystem should be aggregated and linked to build a clinical information commons (CIC) to aid diagnosis
I agree. Saluscoop started as an alternative focused in this approach. As usual, the big question is: who is going to invest in a digital commons?. Unless governments take this initiative as a whole, the future of a data driven medicine is uncertain.



Adrian Piper: A Synthesis of Intuitions, 1965–2016
MoMA, New York, New York

Sat 31 Mar 2018 to Sun 22 Jul 2018

24 d’abril 2018

Equity and QALYs, terra ignota

Incorporating equity in economic evaluations: a multi-attribute equity state approach

Ptolemy used the term terra ignota for regions that have not been mapped or documented. QALYs were born for maximizing health, without any distributive considerations. All the efforts to introduce equity in QALYs have failed up to now. The cartography of QALYs has a pending dimension.
Maybe this dimension is not possible to be defined under a technical perspective, its a societal and policy issue. And at this level decisions are difficult to take.
Anyway, after reading this article you may reach a similar conclusion than mine, or otherwise you can be optimistic about it. It's up to you.

PS. Today I'll give the kenote speech at Col.legi d'Economistes de Catalunya: "La producció eficient i equitativa de salut".

Ai Weiwei

19 d’abril 2018

Man and machine, sharing the decision making effort

Big Data and Machine Learning in Health Care

From JAMA article
It is perhaps more useful to imagine an algorithm as existing along a continuum between fully human-guided vs fully machine-guided data analysis. To understand the degree to which a predictive or diagnostic algorithm can said to be an instance of machine learning requires understanding how much of its structure or parameters were predetermined by humans. The trade-off between human specification of a predictive algorithm’s properties vs learning those properties from data is what is known as the machine learning spectrum
 Higher placement on the machine learning spectrum does not imply superiority, because different tasks require different levels of human involvement. While algorithms high on the spectrum are often very flexible and can learn many tasks, they are often uninterpretable and function mostly as “black boxes.” In contrast, algorithms lower on the spectrum often produce outputs that are easier for humans to understand and interpret.