“Structuring medical knowledge is not even that difficult”
Big data potential and data protection are not necessarily reinforcing each other. Especially in medtech, the vast promises of big data and data science are significantly constrained by patient data protection clauses. As well founded and ethically necessary as this protection might be, the potential of patient data for societal improvement is also unarguably huge. Bypassing insurance companies, and healthcare operator start-ups like Patientslikeme have built up their own databases, making health data of millions of people accessible for broader research. Medexo, a German start-up, is on a similar mission: providing automated and tailored medical advice for patients. We talked to Dr. Jan-Christoph Loh, Founder and CEO of Medexo, about the future of medical advice and potential roadblocks.
Medexo – the easy way to getting a second medical opinion
Insurance companies do have a vital interest that their clients stay healthy. Thus, getting them access to the right therapy in case of illness is essential. However, doctoral assessments are often diverging, making patients inclined to consult a second, or even third doctor. This is where Medexo comes into play: “We help patients get access to a second doctoral opinion of a recognized physician, and in many cases all costs are covered by insurance companies”, explains Loh. Insurance companies cover the costs as second opinions oftentimes reveal that surgery is not necessary, thus in turn helping insurance companies to save significant costs. For doctors, however, this process might not be extremely lucrative. “To provide a thorough second opinion, doctors need to scan through various documents, it’s oftentimes a very lengthy process”, explains Loh. “With Medexo we have also developed a software, which allows doctors to significantly streamline this process.” Medexo’s mission: Providing each patient with the maximum amount of transparency regarding suitable therapies to enable informed decision making.
Highly protected, sensible patient data as key to valuable insights
Key to this mission is essentially a structured database regarding patient history, health metrics and medical records. The patient’s data is however the most protected. Insurance companies, which do have insights into healthcare transactions are not allowed to utilize this data for analysis. And patients themselves are, understandably, quite reluctant to upload their data to a platform. “Whether someone is inclined to upload their medical data is a function of the acuteness of suffering and receival of potential benefits. But especially in Germany, people are not really motivated to make their medical files available. “The incentive has to be quite strong, otherwise people will reject your offer”, according to Dr. Loh. This had significant impact on the business model of Medexo: “Initially we experimented with a B2C model, assuming that people might actually pay to get access to a qualified second medical opinion. But that was not the case”, reflects Loh. Eventually, Medexo’s offering only compelled to users, when insurance companies covered the entire costs and patients only “paid” with their data.
What therapy options fit the medical findings? The complexity of structuring medical knowledge
As Medexo’s platform works parallel to the traditional patient records, it has access to a unique time series of data regarding medical condition, symptoms, treatment and results. “Pre-consultation, we use a questionnaire, which is the basis for the compilation of a ‘self-assessment score’. The same questionnaire is used post-consultation in multiple re-assessments for the coming two years. With this we can evaluate the benefit of our second opinions.”, explains Loh. The challenge, however, remains to construct valid connections between medical findings and adequate therapies. Loh displays optimism: “Interestingly, structuring medical knowledge is not even that difficult. There are many textbooks and other sources, with all the knowledge. At the end of the day, doctors do just the same. They take up the relevant data, process this with knowledge from studies and medical books and arrive at a diagnosis. This can certainly also be automated.” This would open up the field for new and exciting data science opportunities. Due to vast amounts of medical knowledge, though, the starting point was just one discipline (orthopedics), but this approach could certainly be rolled out to most fields of medicine. This, however, takes time: for building a minimum viable product in just one discipline, it took the Medexo team more than half a year.
The vision: accessible medical knowledge, automation of medical consultation
However, the vision is highly appealing: automated medical advice and consultation, allowing patients to have the maximum amount of information handily available for better decision making. Because, as much as you want to prescribe a therapy (imagine a scenario where you could reliably attribute therapy success probabilities), the final decision always has to be called by the patient. “With our vision, we are more or less competing with IBM Watson and Google”, Loh states, “but as opposed to them we have built the infrastructure for a data generating process.”
While the advent of statistics and data science in medicine has born first fruits, there it is still a long way to unlock its full potential. While automated medical consultation might be on the horizon, the role of the doctor as a trusted advisor will most certainly remain crucial for the next years to come. “We are looking more towards developing a decision aid tool for supporting doctors and patients, not towards erasing the role of the doctor”, explains Loh. Even though large groups of people are prone to benefit from these new insights, willingness to contribute data is still at low levels – at least in Germany. The full potential will probably be gradually unlocked, as we further advance our relationship towards data, keeping in mind and balancing the enormous potential it bears for our wellbeing with ethical and data security concerns.