The Congress Center Basel sits right in the heart of the Swiss city, a vibrant meeting point that acted as the perfect venue for this month’s BioData World Conference 2021. Along with a team of Cambridge Consultants colleagues, I was thrilled to be there in person – especially given the optimistic and visionary note struck by Moritz Hartmann of Roche in his plenary talk, which looked forward to more effective data use to improve the lives of millions of people around the world.
Want to discuss further
“Imagine a future where longitudinal data connected across systems leads to improved population health,” was his opening line, which he followed with ways to overcome the obstacles involved in this opportunity to unlock a complex healthcare ecosystem. The challenges include the integration of data across diagnostic modalities, medical and healthcare records. Roche Diagnostics is aiming to manage data across home, care and lab settings and has created a data system to do just this.
Their vision is that this will improve awareness of disease diagnosis, treatment and wellness of patients, with the goal of putting the patient at the centre – resulting in more personalised care. Mapping of multidata sources hasn’t improved significantly over the last decade and one of the main issues lies in government legislation around the sharing of patient data. To obtain patient data, the healthcare industry needs to be able to demonstrate benefits to the patient, and right now there is plenty of effort focused on navigating the privacy and legal challenges.
This was echoed by the next speaker, Michael Seewald, of Astra Zeneca, who raised the example of cardiopulmonary disease in UK. Depending on where you live, there’s been a ten year difference in life expectancy with this condition – in a country with a modern digitised healthcare system. This sort of situation can arise, he said, when healthcare guidelines take years to be implemented.
Citing another example, Michael talked about the treatment of end-stage renal disease, which had suffered from poor diagnosis and a lack of effective treatments in the past. These days effective treatments are available, but patients are not receiving support which could positively impact their health because of the legacy of poor diagnosis. That problem hasn’t gone away.
A data collection ecosystem
All this highlights several issues around the patient care pathway. It is clear that the big pharma companies see opportunities for a significant step change in the way patients should be treated and are calling for radical change. Beyond this, there is a growing awareness that data collection can be used in the diagnosis and treatment of a patient to form part of an ecosystem. Here, big pharma can provide not only the drugs but the framework for the whole patient care pathway – from education, monitoring and early diagnosis through to effective treatment, culminating in what is being called personalised medicine. It is clear that data is the cornerstone of the whole ecosystem. If done correctly, and on a global basis, with deep understanding of the local reimbursement policies and healthcare systems, perhaps big pharma, with its global presence, is indeed best placed to support these initiatives.
With data underpinning these changes, it came as no surprise that the companies helping to make it accessible had a strong presence and voice at the conference. Current healthcare industry challenges include how to protect patient privacy and maintain regulatory standards, encompassing legal and compliance risks as well as cyber security issues and beyond. According to the TripleBlind CEO, Riddhiman Das, 43 zettabytes of data are currently stored by enterprises, which are inaccessible and not commercialised. The primary reasons are privacy and regulatory concerns.
The industry accepts that it cannot send data sets to other companies to give them replicated data at their site. Once the data is given away in this manner, it can be used in other unintended ways. For example, data used for prostate cancer diagnosis could be used to determine ancestry or male pattern baldness and much more. Companies such as TripleBlind sit in the middle, allowing company A to share its data with company B, without raw data ever leaving company A, and company A not being able to see or access company B’s algorithms. One way encryption is used, which allows data to only be analysed by one algorithm. It cannot be used for other tests and thanks to encryption the algorithm is not divulged to the other party.
This has utility beyond sharing of just genetic data, allowing access to multiple data types from different sources for AI analysis to early indication reporting from clinical trials before completion of the trial. It is very clear that the requirement of a collaborative healthcare ecosystem opens a plethora of opportunities to use data, not only to monitor and influence clinical outputs, but to feed it back into R&D efforts.
Making the most of data
Once we have overcome the challenges of sharing data through the ecosystem, ensuring robust privacy and regulatory policies are in place, how do we make the most of it? It’s clear that AI is increasingly being applied by software providers as well as big pharma across much broader applications, and it is starting to bear fruit. The amount of data being generated, from patients and from basic research is so significant that it has prevented it being analysed alongside data sets from other modalities, but that also seems to be changing rapidly.
Prioritising suitable assay models and appropriate chemical libraries has always been a challenge for pharma. Screening millions of compounds, through several primary screens, takes a very long time and comes at a significant cost. AI means that companies are now becoming more strategic in how they apply these screens. Edmund Champness of Optibrium described how they are helping the likes of Genentech to utilise AI in day-to-day drug discovery, with an example of kinase profiling. They use their proprietary AI model, which is able to take existing data points and predict with levels of confidence virtual compound activity in variable experimental models, thus doing some of the primary screening virtually. This reduces time and cost of the development cycle. This really does feel like the future of drug screening.
In another presentation, Christopher Miller of GSK discussed greater efforts to start using multi-omic data more effectively. From being considered a fad by some in the earlier days, he says that “single cell -omics is revolutionising biology”. Functional genomics is where variants in sequence are mapped to changes in cellular function. Across multimodalities, there are vast quantities of data. Only by being able to utilise AI and ML, are they are able to see patterns in very complex loci of genes.
This data can be fed into an AI model to rank genes, which guide functional genomic studies leading to further predictions. AI is not only analysing results but is also making predictions with this data being fed back into the system to develop iterative learning. In some instances, the results are non-intuitive, so are unlikely to be designed by a human, but they have been used to drive future studies and these are starting to prove extremely valuable.
Ultimately, greater access to patient data, and the subsequent analysis of the data using AI could transform the healthcare ecosystem towards democratised access to diagnosis. In turn, healthcare professionals would gain new testing options, giving increased accuracy and faster time to results and treatment decisions. For patients it would mean better access to precision testing and affordable treatments. My hope is that we’ll see improved data sharing in order to democratise disease diagnosis, treatment and wellness of patients. Focusing on the patient-centric goal will lead to more personalised care, globally. Maybe the big pharma companies are best placed to orchestrate this, by nurturing a fair and collaborative sharing and utilisation of data.
Please email me if you’d like to share your view on the topic, it would be great to hear from you.
