Can AI-driven digital biomarkers redefine diagnostics and clinical trials?

Artificial Intelligence and advanced analytics technologies have huge potential to improve the world of health. We talked to Teva employees who are using the technologies.

Artificial intelligence (AI) and advanced analytics technologies have the potential to improve almost all aspects of human life.

The AI technologies, which mimic human intelligence and enable machines to autonomously or semi-autonomously analyze large amounts of data, are becoming more common in healthcare and the pharmaceutical industry. In particular, digital health technologies are increasingly being used as endpoints in complex neurological indications, creating sensitive, objective and non-invasive neurological measures, ensuring confidence and scalability with emerging digital technologies.

These technologies, for example, are already being used in some countries to improve the speed and accuracy of diagnosis and screening for diseases; to assist with clinical care; strengthen health research and medicines development, and support diverse public health interventions, such as disease surveillance, outbreak response, and health systems management.

The global market for AI in healthcare may be worth $61.59 billion by 2027, up from $3.39 billion in 2019, according to forecasts.

Teva is using AI and advanced analytics to develop digital biomarkers for several conditions including schizophrenia and neurodegenerative diseases (ND), including Parkinson’s disease (PD) and Tardive Dyskinesia (TD), and improve clinical trial processes to get medicines to patients faster.

Digital biomarkers

According to the National Library of Medicine, a digital biomarker is defined as a characteristic or set of characteristics collected from digital health technologies that is measured as an indicator of normal biological processes, pathogenic processes, or responses to an exposure or intervention, including therapeutic interventions.

Or to put it another way, digital biomarkers are tools that can identify diseases and measure their progression, objectively and sensitively, as well as monitor how patients are living with the condition. While clinical – human-based – assessments can be subjective and costly, using digital biomarkers to assess conditions may be less costly and produce more consistent results.

Teva is involved in a project called MOBILISE-D, which aims to develop a comprehensive system to analyze people’s gait based on digital technologies, including sensors worn on the body. It focuses on conditions which often affect mobility including chronic obstructive pulmonary disease (COPD), Parkinson’s disease (PD) and multiple sclerosis.

Working with teams of clinicians and biomechanical engineers, Teva is developing technology for effective patient assessment. Patients with PD wear sensors on their lower back, ankles and wrists and the sensors' data is analyzed to depict walking speed, regularity and symmetry, among other factors.

“We are using AI and analytics to try to predict and assess the motor stage of the patients' disease, which affects the body’s movement,” says Michal Melamed, Data Scientist, Advanced Analytics and AI at Teva. “And to identify the walking factors most associated with Parkinson's disease motor stages.”

“The goal is to connect digital mobility tech with clinical outcomes,” says Michal. “Our hope is to gain regulatory approval of the wearable technology and data it gathers, which may be used in clinical trials to aid in the understanding of and treatment needed to tackle the progression of Parkinson's disease."

It is also possible, explains Michael Reich, Teva’s Director of Advanced Analytics and Artificial Intelligence, to use facial recognition tools to identify symptoms of TD, for example. TD is a condition characterized by involuntary repetitive muscle movements and is often misdiagnosed. Digital solutions can help in the detection, assessment and monitoring of TD – and can involve using something as simple as the patient’s own mobile phone to identify symptoms.

Teva is currently working on AI-related algorithms to extract facial landmarks from a mobile phone video to identify the symptoms related to TD. The assessment can be done remotely, does not require the direct intervention of a specialist doctor, is completely objective and potentially more sensitive, as it will detect even minor changes that might evade a human rater, providing detailed analysis rather than observations.

Voice analysis is another area of digital interest, as speaking is a complex process that requires coordination of cognitive, muscular and respiratory systems. Analytics & Big Data expert Alex Gotler is leading the voice analysis project at Teva.

Using mobile phone voice recordings, features such as speech rate, emotion, vowel variability, timing variability and vocal resonances can be measured. Voice analysis is used in many psychiatry domains including Parkinson’s disease, asthma/COPD, coronary artery disease and schizophrenia.

“We are using voice recordings to extract key features from patients’ voices and developing models to correlate them with the Positive and Negative Syndrome Scale (PANSS), the gold standard scale for quantifying the severity of symptoms in schizophrenia,” Alex explains.

Clinical trials

NDs are a common and growing cause of morbidity worldwide and developing medicines to treat these conditions requires the participation of large numbers of patients in clinical trials for long periods of time. Using digital biomarkers may double the success rate of trials, reduce the patient numbers needed and shorten the length of the trial period.

Recent research shows the probability of success for new drugs with and without biomarkers in the US through the various stages of development, in the period from January 1, 2011, to November 30, 2020. It was found that while the probability that a new drug without biomarkers makes it from phase I to approval was only 7.6 percent, those with selection biomarkers had a probability of success of 15.9 percent.

It is important, says Alex, to stress the challenges of running clinical trials in terms of the expense involved and the cost of trial delays in lost revenue and delayed timelines to market. Optimizing clinical trial site selection procedures may accelerate performance and Teva is currently using vast stocks of data to analyze previous site experience and profile sites that are more likely to recruit on time and have access to the required patient groups. Optimized performance may result in a more efficient patient study and improve the time to market for important medicines.

Advanced analytics and Artificial Intelligence have huge potential for the future of health care. At Teva, we continue to explore the possibilities that may help to improve the lives of patients worldwide.

NPS-ALL-NP-01226-MARCH-2024


Find out more

Share this article:

You might also be interested in

Celebrating Pride, Creating Connections

Celebrating Pride, Creating Connections

Challenging Gender Stereotypes to Drive Excellence

Challenging Gender Stereotypes to Drive Excellence

Teva’s Top Graduates: From Academic Learning to Business Leadership

Teva’s Top Graduates: From Academic Learning to Business Leadership

Breaking Boundaries: How Diversity Fuels Innovation in Teva’s R&D Management Team

Breaking Boundaries: How Diversity Fuels Innovation in Teva’s R&D Management Team