Digital health is exciting, but it also has life-threatening bugs

31 January 2022

Digital health creates an opportunity for better distribution of health services, higher quality of care, and a more substantial role for chronic diseases prevention. However, in your book, you claim that digital health is risky. Where does this standpoint come from?

My interest in healthcare computing started when one of my students ended up in intensive care. I saw some warning notes on one of the infusion pumps, and I decided to buy one with a research grant I had at the time.

As a professor of computer science, I was horrified at the low quality of the infusion pump, but maybe I’d misunderstood its clinical use? So I talked to an anaesthetist, “went undercover” for a few days, scrubbed up, and “helped” in several operations.

In all of the operations, something digital went wrong. The skilled teams recovered in each case, but there should never have been problems in the first place. For example, one day, a ventilator crashed and had to be rebooted.

At the beginning of the book, you write: “healthcare is being held back because we don’t see computer’s bugs, and we don’t understand risks caused by them.” Or maybe we simply don’t have a better alternative, and mistakes in digital healthcare must be included as an “acceptable risk”?

It’s interesting that the same comment for a different area would be met with astonishment! “Nuclear power is being held back because we don’t see computer’s bugs, and we don’t understand risks caused by them.” Note that far fewer people have died from software bugs in the nuclear industry than in healthcare. Healthcare is an unnecessarily dangerous place.

You have even developed an interesting “cat thinking” theory – people are so attracted to digital innovations that we are blind to their risks. Is it also the case in healthcare where sometimes I have an opposite impression healthcare is a highly regulated and conservative sector and therefore lags behind digitalization in other sectors.

My experience is that healthcare’s regulation of conventional healthcare (e.g., drug trials) is rigorous, but healthcare hardly pays any attention to safe digital because digital is seen as a saviour. 

Cat thinking is introduced with a story about my own cat — how his purring releases hormones so we feel happy. Once we feel happy because of hormones, rigorous thinking is irrelevant. And much digital is designed to make us happy — so we want to buy it and so it sells well — rather than work well so it’s safe.

Can you then give some examples of the risks of the digitization of hospitals?

My book discusses 73 nurses who were disciplined for all making the same error. It turned out that it was a computer problem. I also discuss Olivia Saldana, who is in prison for manslaughter — because of computer bugs. And Kimberley Hiatt who committed suicide after a calculation error. And numerous Down Syndrome errors thanks to Y2K. And many more stories!

You claim that digital health is often misapplied. What do you mean by that?

Excitement with AI might be a good example. Of course, AI can work, but it isn’t a panacea. One example is when a company offered AI to a hospital, took their data, then the company was sold on — the result was that patient data got lost to the new company who hadn’t signed the original agreements.

So let’s wrap up: Where do the main risks come from, and can they be prevented, assuming that we are only in the early stages of digitization of healthcare and such risks are almost inherent in the development of new technologies? On top of that, the alternative “analogue” interventions used so far are not entirely safe either...

Poor regulation is the key problem. Manufacturers/developers are not required to be competent in any sense of the word, and they aren’t. In most countries, it is illegal to have a car on the road if it hasn’t passed appropriate safety tests, either during manufacture or periodically afterward (every year in the UK). There is no equivalent for any healthcare digital, and of course, it is unsafe as nobody checks it!

You dedicate the last part of the book to a “better future.” What is the actual digital promise, and what needs to be done to make it happen as soon as possible?

Well, of course, digital is amazing and is doing great things. But, for example, the skills Apple applies to make iPhones unique is 10,000s of times more advanced than the skills typical digital healthcare companies (let alone hospitals) have access to. Concrete example: one app I discuss in the book had a student programmer. Apple has over 1,000 people doing QA in just one of their many buildings on one of their many campuses.

Aren’t you afraid that the book will become grist for the mill of the septics of digitalization and, at the same time, will be ignored by the supporters as biased?

The book is full of human stories, and the stories are all carefully documented. Of course, it’s biased in that the majority of the stories are disasters. Still, when you read “success” stories of digital elsewhere, there is very little evidence that they are really successful — it’s almost all triumph of hope and promises over evidence.