On average, humans have one testicle

Joan Escarrabill

This article deals with the interpretation of data. The title is an example made by Daniel Levitin in his suggestive book Weaponized Lies: How to Think Critically in the Post-Truth Era (Penguin Random House, New York, 2017), about the mistakes that can be made and the lies that can be said according to how numerical data is exposed. Obviously, the title of the article is exact. If the proportion of men and women is about 50%, certainly humans, on average, we have a testicle (and an ovary).

The confusion that can be created if we misinterpret the relative risk is well known, but when we refer to the data, there it’s not only an interpretation problem. Now, with the huge accumulation of data we have, we take another leap: data analysis it’s not only interpretation (talking about clusters is a way to interpret data), but, by means of algorithms, it can lead to prescription in an automatic and personalized way.

I just read Krisa Taylor's book The Patient Revolution. How Big Data and Analytics Are Transforming the Health Care Experience (John Wiley & Sons, Hoboken, New Jersey, 2016). The author states that the great challenge we have now is the large volume of data we can count on. We no longer only talk about the clinical history data. Now we can easily integrate the data of wearables (which collect passively, without any direct intervention by the patient) and the data of apps, which can be passive (geo-location, for example) or active (the patient answers questions). Finally, there is also some data that automatically generates the devices. You don’t have to have a lot of imagination to think about other available data that can be integrated (with good intentions) to improve care. Without going any further, the electricity consumption can give a lot of information that hypothetically can have clinical uses (I make a hypothesis: knowing how many times an elderly person who lives alone can wake up at night can alert about the risk of falls); or all the contacts that a patient has had with assistive devices (health, community pharmacy, rehabilitation centres ...).

Taylor exposes the hypothetical benefits of integrating all this data in the cloud. It's what she calls the Personal Health Cloud (PHC); the PHC, on the contrary that most of the current instruments, guides our behaviour, instead of following it.

The automatic analysis of personal data (personal health analytics) from algorithms (deep learning algorithms) can generate recommendations or alarms before a potentially dangerous event occurs. In the case of chronic diseases all this information can generate algorithms that are a good tool to change lifestyles (diet, physical activity, healthy sleep, adherence to treatment, etc.). B. J. Fogg spoke about the concept of persuasion through technology more than twenty years ago, and in 2015 the MIT Technology Review published a monograph with the title "Persuasive Technology".

We are in the beginning and, therefore, we know very little of all this. It it’s not necessary to have to be especially lucid to foresee that the management of all this information poses ethical problems. Some of the problems are related to the fact that the data is located in multiple places (sometimes in places unknown to the user): algorithms that optimize tasks are not the same as algorithms that make decisions. Or it’s important to know to what extent the algorithms make decisions without human control (transparency of the algorithm). Algorithms can be good at predicting results, but predictions are not causes, and much more.

Prior to this new world we can take two positions. One is the precautionary principle: as we don’t know the results (current and future), we must avoid action in this are until we have more knowledge. It seems a prudent attitude. But it raises two problems: first, the data is already in the cloud, surrounding and affecting us, and secondly, is it ethical not to use available data that can generate immediate benefits for citizens?

Another position can be a kind of principle of exploration. The serious explorer poses challenges (difficult) but it’s not suicidal. The explorer is tenacious and tries to achieve his goals in different ways, until he succeeds.

In any case, what it’s not worth is to act as if nothing happened.

As in each article, I propose a piece of music. I have chosen the Chaconne by J. S. Bach (Partita for violin only number 2 in D minor, BWV 1004). It’s a small tribute to Jorge Wagensberg (1948-2018). Jorge said that the first thing the children had to learn was language: oral and written (in three languages), the mathematical language and the musical language (perhaps he also said the artistic language, but I'm not sure). Bach is the most brilliant representation of musical language. Thank you, Jorge, for making us understand the importance of language, that questions are always more important than answers, and that it’s always more useful to doubt than to have faith.