Close
Close

Published

AI feeds booming scientific wellness sector, as well as its hype

Even without AI, Scientific Wellness has long been a booming business – one often mired in controversy. It has been derided on the one hand as a first world luxury, while also being welcomed as a major extension of healthcare by helping people to stop becoming ill in the first place. It is a very broad sector as can be seen by considering the highly diverse applications it embodies, as well as the colossal market predictions.

According to the Global Wellness Institute (GWI), a non-profit industry body, the global wellness market grew 10.6% to $3.72 trillion from 2013 to 2015, the latest year for which data is available, over a period when the global economy shrank 3.6 percent. Not surprisingly, the US has led this field, ranking number one for spending in four of the five wellness categories tracked by GWI –  spas, workplace wellness, wellness tourism, and wellness real estate. As this list suggests, the field extends from the fitness and luxury health sectors at one end, to more directly healthcare related applications like physiotherapy.

AI has emerged with the promise of revolutionizing this sector like others and also to put it on a more rigorous footing. However, in the immediate term the arrival of AI has been a mixed blessing because it has also boosted the less reputable parts of the sector and been recruited by charlatans peddling pseudoscience with claims that it can generate cures for cosmetic problems, such as creams for skin conditions like acne, without any real evidence they work.

An underlying problem is that scientific wellness is a field of promise whose applications and benefits lie mainly in the future, and so is still riding the hype curve. The main scope of AI as generally in healthcare is in detecting patterns in data sets that help diagnose existing conditions, predict disease developing in future or prescribe appropriate treatments. It holds great promise for personalized medicine by combining genetic profiles with diagnostic data to recommend treatments suitable for a given patient that might not work for someone else with the same underlying condition.

In principle, it has already been established that analytical techniques can diagnose a wide range of conditions through application to metabolic data sets including variables such as blood temperature, pressure and glucose levels, as well as numerous other biomarkers including the amino acids, constituents of proteins, fatty acids and nutrients. By observing levels of these and their variations across a short period of time, many deductions can be made accurately, ranging from whether the subject has just eaten and what, to early stages of different diseases.

The problem is the skills and algorithms for deriving inferences from complex metabolic data sets are in their infancy, and as a result, claims of benefits made by many parties with commercial interests do not stand up. Unfortunately, some eminent scientists have been associated with at best debatable projects or start-ups on the scientific wellness front, one being Leroy Hood, a pioneer of automation in the laboratory and one of the fathers of systems biology, which sought to turn life sciences into a more predictive field based on mathematical modelling.

Now 79, Hood has taken to scientific wellness over the last three years with a vision of preventing disease and he began with a pilot study designed to demonstrate the analytical potential of large metabolic and genetic data sets. He created what he called “personal, dense, dynamic data clouds” comprising data on 108 people including levels of 643 metabolites and 262 proteins combined with their full genome sequence and information about sleep patterns as well as physical activity, all plotted over 9 months. This led to various claims of benefits including warning of early stage type 2 diabetes with regimes to alleviate it, including exercise and diet. This spurred him to co-found a company called Arivale, which now offers “wellness services” including monthly coaching based on such analysis, charging $3,499 for a first-year membership.

The problem can be seen from the blurb, which states that the company offers “a scientific path to optimize wellness and avoid diseases.” This it says is based on analyzing various areas of an individual body, such as genome, blood and saliva, gut microbiome, and lifestyle, so that it can provide actionable recommendations for improving wellness. It also boasts that it can personalize recommendations on diet and exercise to individuals based on genetic profiles and other data.

That is just jumping the gun before anywhere near enough research has been done to establish a clear evidence-based foundation for such claims, which is some years away except in some isolated cases. It is no wonder some of Hood’s colleagues have expressed reservations about this service, although there are plenty of other scientists who have been seduced prematurely by the wellness dream – and the potential here for AI.

Yet within more limited ambitions scientific wellness is already delivering tangible benefits, having attracted interest both from major AI firms and various startups. IBM has been quite active on the wellness front and has been collaborating for example with Technogym, provider of gym equipment from exercise bikes to free-weights, and which now calls itself a wellness specialist, to develop a “virtual coach”.

This applies IBM Watson’s cognitive computing technology to crunch data obtained from some of the gym equipment as people use it and combine that with health conditions, diet, weather and other factors to recommend a personalized training program. At least this comes with no overtly exaggerated claims and yet can apply advanced techniques to improve recommendations, even if there is no absolute proof of efficacy.

There are also wellness technologies closer to the healthcare sector offering tangible benefits. One example is the New Jersey based start-up Stages.co, which has applied AI to augmented hearing through headphones by filtering out unwanted ambient sound and distractions. This combines signal and speech processing with a custom circular microphone array to focus on desired incoming sounds or voices according to user preferences. It has been shown to be particularly helpful for people with impaired hearing in places such as restaurants where there is a lot of background noise. Effectively users can tune in on the sound they want to hear like while others fade away.

Defining such a huge range of applications under the umbrella of scientific wellness makes the field hard to analyze methodically. However, this last example of augmented hearing suggests that the field can be broken down between such embedded specific applications where the intelligence resides in signal processing and others that apply pattern recognition and training. This second group can in turn be broken down between applications that are purely diagnostic or prescriptive of conditions that may emerge in future, veering more towards healthcare, and those that suggest personalized programs to boost quality of life or wellbeing.

Close