A data-driven approach: Harnessing the value of omni data to address progression in neurological diseases

Gaining a deeper understanding of disease progression in neurological disorders has long been my passion and is also a shared purpose for us working in neuroscience at Roche. When I worked as a neurologist, I saw first-hand the devastating impact of disease progression – whether that is losing independence with the ability to walk or work, or no longer being able to recognise loved ones. As a neuroscience community, it is imperative that we find ways to better understand the course of a disease and predict when and how progression will occur. Not only will this help us to support people with neurological disorders – helping them to make more informed choices about their future, but also to make sure that our healthcare systems are better equipped to tackle the growing prevalence of neurological disorders.?

I am pleased to say that in the last few years we have seen progress, a lot of which can be credited towards access to different sources of high-quality and comprehensive data – sometimes known as ‘omni data’. From artificial intelligence to real-world evidence and human data, together these data have allowed us to gather insights that were previously unattainable, at a scale and speed that were previously out of reach.

Here are just some of the ways that data is informing progression in neurological disorders.

1. Human data: Providing our guiding light

Human data accounts for what defines us as people and is at the heart of a patient centric approach to healthcare.

In spinal muscular atrophy (SMA), it's generally accepted that the earlier treatment starts, the better the potential outcomes are in younger patients. However, in older patients, the goal of treatment is to prevent progression and to stabilise the disease. This nuance means that we must collect different types of human data for each patient group, and recognise that at different ages, different outcomes will be most important. In clinical trials for younger patients, we may look at a baby’s ability to sit for five seconds without support, turn over, hold their head, stand, and walk. Yet in adults, we need to collect data that focuses on maintaining motor function, against baseline.

2. Digital biomarkers: Alerting us to disease progression through our mobile devices

Digital biomarkers are physiological and behavioural characteristics that are collected and measured by digital devices.[1] At Roche, we are collaborating with the Michael J. Fox Foundation for Parkinson’s Research on the Parkinson’s Progression Markers Initiative (PPMI) study. This is a digital biomarker platform that uses smartphones and smartwatches to monitor posture, coordination, and movement through a series of daily tests assessing motor symptoms, such as speed of finger movements, and balance.[2] Initiated in 2018, the database is uniquely placed to make contributions to the broader research field and provide us with real-time insights into Parkinson’s symptom progression.

More recently, we joined a global collaboration with the Digital Medicine Society and other industry players, to select and develop the most effective digital biomarkers for people living with Alzheimer’s disease. Data will be collected from wearable devices or tests, such as analysing voice recordings or hand movements. This could help us more accurately measure disease progression, improve care management and inform innovation to help advance the science of this devastating disease.[3]

3. Artificial intelligence: Predicting the future

As many of you know, artificial intelligence (AI) is the simulation of human intelligence processes by machines. The progress that we have seen in AI over the past few years has been considerable, and the same can be said for its application in neurological disorders.

Research from a collaboration between Fujifilm and the National Center of Neurology and Psychiatry (NCNP) suggests that AI technology could predict whether a patient with mild cognitive impairment will progress to dementia within two years with an accuracy of up to 88%.[4] These impressive results have been achieved by using advanced image recognition to monitor the progression of Alzheimer's disease from three-dimensional MRI scans of the brain. In combination with deep learning, AI technology that monitors the hippocampus and the anterior temporal lobe - two areas highly associated with the progression of Alzheimer's disease - fine atrophy patterns associated with Alzheimer's disease, can be detected.[3]

4. Real-world data: Ability to examine progression

Real-world data can be derived from a number of sources, including observational studies and patient registries, but excludes data from controlled clinical trials.

For a disease such as MS, where there is a high level of heterogeneity between patients’ experiences of disease pathology and progression, real-world data is essential to improve our understanding and inform more personalised treatment approaches. Analysing real-world data gives us access to large patient cohorts and allows us to examine different theories about progression in the presence of a diverse patient group, regardless of treatments used.

5. Big data: Collaborative data sharing

Referring to varied, voluminous sets of data from medical and hospital records[5], big data supports research and the development of biomarkers to increase our understanding of neurological conditions.

Instead of using the “traditional” centralised data-sharing networks, whereby all data is sent to one central location, we are advancing federated architecture learning, a technical and mathematical model, where data is stored where they are owned (e.g. smart devices) or produced (e.g. an imaging centre). With this new model, the hope is to reduce the complexity of data exchange and promote more efficient, disease-area focused data while maintaining privacy. This will support sharing data between organisations and enable better understanding of conditions such as MS.?

By expanding the use of an ‘omni data’ approach, I firmly believe that our industry will be able to deliver more treatments, at a lower cost to society – a commitment that we, at Roche, are striving towards. High quality data can increase R&D efficiency and the chances of success of our clinical trial programmes. Not only do we get more frequent data, but the data is ‘closer’ to the patient experience. With this information we will be able to better demonstrate the value of potential treatments and management approaches to healthcare systems and society, which in turn could lead to more access to life-changing medicines for all.

References

[1] Karger. Digital biomarkers. [Online]. Available at: https://www.karger.com/Journal/guidelines/271954#:~:text=Digital%20biomarkers%20are%20defined%20as,or%20predict%20health%2Drelated%20outcomes. Last accessed: July 2022

[2] Parkinson’s Progression Markers Initiative. About PPMI. [Online]. Available at: https://www.ppmi-info.org/about-ppmi Last accessed: July 2022

[3] Digital Medicine Society. Digital Measures Development: Alzheimer’s and related dementias. [Online]. Available at: https://www.dimesociety.org/tours-of-duty/digital-measures-adrd/ Last accessed: July 2022

[4] Fujifilm. Fujifilm and National Center of Neurology and Psychiatry announce positive study results of Fujifilm’s AI technology predicting mild cognitive impairment conversion to Alzheimer’s disease. [Online] Available at: https://www.fujifilm.com/jp/en/news/hq/7926 Last accessed: July 2022.

[5] OCI. What is Big Data?. [Online]. Available at: https://www.oracle.com/uk/big-data/what-is-big-data/ Last accessed: July 2022?