Healthcare of the Future: Data Driven Personalized Medicine and the Probability of Success

(This post first appeared on Pulse)

As a patient I would like my doctor to be able to not only tap into his or her own experience, but to use the experience of all of his or her colleagues around the world before prescribing me my therapy

Data gets us home faster and wins games

I am a big fan of the traffic app “Waze”. Every morning I leave the house to go to work, I enter the address of my office in the app and it gives me the most optimal route. Now, I know most of the roads leading to my office and on a Sunday morning in July I could probably guess the shortest route myself. But on a dark, rainy morning in November with accidents and road constructions everywhere the probability of me selecting the most optimal route would be very low. In that situation I rely on the advice given and the data generated by my colleague drivers. While they actively report traffic jams, road blocks, etc. the app also monitors the velocity of all the users. Based on this data and associated pattern recognition I get presented with the fastest option. Having experienced the value of this approach (I made flights which otherwise I would have missed for sure), I gladly reciprocate by reporting events on my route as well.

For those of you who have seen the movie Moneyball with Brad Pitt, follow the Golden State Warriors’ “Strength in Numbers” campaign or seen this Ted Talk, you can understand that also in sports, the probability of success increases if you understand the patterns, the numbers and the data.

Towards data driven medicine

Probably one of the most delicate bonds is the relationship between a patient and a doctor. We rely on the physician to help us preserve what is most valuable to us, our health. To make sure the MD can do a good job we expect them to study many years, be an intern for a long time and gain as much experience as possible to be able to select the right treatment for our illness. We also expect them to keep our information strictly confidential and prevent any misuse. When looking at the evolution over the last centuries the medical world has done a terrific job. Life expectancy has drastically increased and many diseases that once were fatal have become curable or very manageable. At the same time, however, it is very clear that the revolution of the last decade, such as the use of big data to detect patterns to help us navigate traffic or help coaches win games, has not yet found its way into healthcare. One reason, could be that adequate use of big data requires collection of high quality data which in turn would imply that the MD shares and centralizes (anonymized) patient information. It seems that doctors in general are reluctant to share patient information due to privacy concerns or because they are afraid to potentially lose the patient to other healthcare systems or providers. And even though these concerns might be valid to a certain extent, the potential upside of data driven, personalized medicine is hard to overstate. Imagine that when you visit a doctor, your case is not just assessed by one physician but is compared to thousands of similar cases around the globe. Imagine instead of having one human interpret the multitude of test results, a computer looks for patterns and based on this can suggest the best course of action. It is not difficult to see that in a situation like that the probability of selecting a successful treatment drastically increases. Will this approach make medical doctors obsolete? Absolutely not! MD’s will continue to manage care but will be able to do this with better tools. An airplane is technically capable of flying without a pilot but the profession of pilot will not disappear anytime soon. In a similar fashion the doctor will remain in control of the patient’s healthcare journey.

My company is very active in respiratory medicine where we study lung diseases, such as COPD, and try to develop better image-based diagnostics. One thing that has become very obvious over the years is that the umbrella definition of, for instance, COPD (which stands for Chronic Obstructive Pulmonary Disease) covers a very broad range of patients with a large level of heterogeneity. Looking back, it can be determined that the lack of good diagnostics and the absence of a pattern recognition approach has led to a disease definition which does not facilitate the development of very effective drugs. The majority of COPD patients is treated with bronchodilators, sometimes in combination with steroids. More and more alternative therapies such as endobronchial valves and coils are studied. The results of these studies showed that these therapies can be very effective for selected patients. To select these patients properly, however, better diagnostics are needed and whenever possible making use of the latest big data approaches. The good news is is that virtually all technical requirements for this novel approach are met and that the implementation is imminent to empower the treating physician.


The cost of personalized medicine

Often we hear that personalized medicine is not affordable. We need the volume of block buster drugs and large markets to keep healthcare costs down. But is that really so? The innovations that have made our lives more productive over the last years, like the traffic apps, are offered either free of charge or at very little cost. The data generated by the user is so valuable that the system can be made sustainable without a large cost to the user. So instead of trying to make a one size fits all drug or treatment that needs to serve a (heterogeneous) patient population, let’s try to turn things around. Why don’t we create a large volume of data and knowledge by bringing together the results of individual patient assessment including knowledge around effective therapies for these specific patients. The resulting database will allow reliable pattern recognition to determine selection criteria for specific therapies leading to better outcomes. For instance in the respiratory space, we will know what type of patient will benefit from coils, valves or surgery [1]. For which patient we will need to consider Non-Invasive Ventilation [2] in addition to the inhalation therapy or when a systemic drug is preferred over or in addition to an inhaled drug [3]. Leveraging the data this way can offset potential additional cost for personalizing medicine: ineffective but often very expensive therapies will be avoided and patients’ quality of life will improve leading to higher contribution and participation in social life. So every personalized assessment done anywhere in the world has the potential of saving or improving the life of any other similar patient elsewhere in the world. Will the implementation be easy? For sure not. Current healthcare reimbursement systems are not yet setup to support this innovation and the medical professionals will need some convincing but is it worth the effort? Absolutely!

Every personalized assessment done anywhere in the world has the potential of saving or improving the life of any other similar patient elsewhere in the world


Sorry, comments are closed for this post.