With both an elite data science degree with distinction from Ludwig Maximilian University of Munich in Germany and full medical training at Harvard Medical School in the US and the Ludwig Maximilian University of Munich in Germany, including a doctorate in neuroscience, Dr Pfister currently leads multiple initiatives in the field of Health AI, developing next-generation diagnostic solutions to improve patient care and enable personalised medicine.
Dr Pfister is also the Co-Founder and CEO of deepc, an artificial intelligence health diagnostics startup.
1. The healthcare industry today is markedly different from that of even a few decades ago. As a clinician and industry expert, what are some of the most significant changes that you have observed particularly in the role of diagnostics?
Advancement in genomics and technology, such as sequencing, in recent years enabled laboratory medicine to develop many high-value tests. These tests can more precisely identify disease, even its subtypes and monitor progression. In the field of oncology, in particular, new biomarkers are telling doctors who would respond to which drug, setting the foundation for personalised healthcare.
At the same time, we are witnessing the early stages of digital biomarker development as the potential new pillar of diagnostics. Especially for the management of chronic diseases and early disease detection, digital biomarkers carry the potential to bring transformational change.
2. How do you see data analytics reshaping the healthcare industry?
There are many areas where I already see data-driven AI solutions bringing tremendous improvements to existing diagnostic pathways. AI and technology have great potential to open up new ways of how we perceive, approach and deliver healthcare in the future.
To address the rapidly rising chronic disease prevalence that is threatening many families and society, new tools are urgently needed.
For example, we are working on an AI solution to provide physicians treating Parkinson’s Disease patients with continuous treatment response insights via motion data that is captured via a wrist-worn sensor. Based on the motor pattern that our AI solution provides, physicians can precisely adjust treatment dosage, frequency and schedule to ensure the Parkinson’s Disease patient stays in optimal therapeutic range and therefore minimize treatment-induced motor fluctuations, which have a significant impact on the functionality and quality of life of patients and pose a serious cost burden on the healthcare system.
Advances in Internet of Things (IoT), data and AI have the great potential to form the newest diagnostics pillar – digital biomarkers – and provide doctors with continuous, objective, real-world insights to optimise chronic disease patient care.
For the first time, we are able to process a tremendous amount of health data.
Accurately equipping this data with clinical information and using it as input for machine learning models can generate unprecedented insights that power well-informed clinical decisions. This will enhance the efficacy of treatment, even to an extent of individually tailored care thereby ensuring a higher quality of life; improve efficiency in healthcare delivery, and ultimately prevent diseases.
3. What sort of data structure best suits our needs today?
Accurate AI stems from a large amount of structured data. But, it is not only about the quantity. Every single human is unique: their genetic profile, their phenotype and biometrics, their habits, their disease conditions, their geography, so on and so forth.
No institution in the world can or will be able to cover this need for data diversity alone and data sharing, at a large scale, is inevitable.
We see how today’s healthcare data silos impede the world’s defense against the COVID-19 pandemic that is hitting us hard at present. Due to the sensitive nature of healthcare data, privacy is a key concern and must be preserved. At the same time, the desire for each party to retain data control and ownership can increase the friction in data sharing.
Ultimately, collaboration will only be achieved by changing our mindset.
From an infrastructure perspective, moving from a centralised data silo structure to decentralised systems could be a game changer. Federated learning allows AI algorithms to travel to datasets for training rather than pooling the data into a single location. Data stays at its own location and therefore, privacy and control can be ensured. Also when you think about the need for massive data storage capacity, this concept makes a lot of sense. Better to let an algorithm travel to the data for training than shipping tons of data from A to B. The caveat is data harmonization – the AI algorithm needs to learn from datasets in the same format, standard, etc. I’m hopeful with new technologies that tackle the frictions of data sharing, there will be more collaboration in the healthcare space in the near future.
4. What do you think the future of a data-powered health ecosystem will look like, particularly in Asia Pacific?
Many emerging markets in Asia Pacific have the opportunity to leapfrog as they exhibit lower barriers for health tech integration with fewer legacy systems. As a result, they can shape the way healthcare will be delivered as they develop their healthcare ecosystem.
In certain disease areas, individual ownership and management of data will increase, such as in the chronic disease space where prevention can make a huge difference.
Overall, a successful healthcare model will be a hybrid of both personal and institutional management of disease and in both cases, data will be a key enabler for cost-effective healthcare delivery.
With the increased control of data privacy corresponding to each jurisdiction within the region, decentralised systems with unified interface and data structures will be the ultimate propeller for a productive data-powered health ecosystem to benefit people and the society.
At the same time, the fragmented landscape of data-driven solutions needs to be consolidated in the mid-term. As clinicians and healthcare providers recognise the power of data and digital biomarkers in disease management, we may see the emergence of platform providers offering an end-to-end service to bridge the gap between digital biomarker innovation and broad clinical adoption and deliver holistic value.
Read more about digital technology in healthcare at Roche diagnostics news.
For more related articles: Building A Data Ecosystem For Better Healthcare, Towards A More Human Lab, Moving Diagnostics And Medicine Onto The Technology Curve
