As the cost of developing new therapies continues to rise, pharma executives must turn to technologies that drive better insights into de-risked investments and produce operational efficiencies. AI-powered digital measures are being developed and validated to better define disease, patients, and drug efficacy.
Drug development represents one of healthcare’s most significant challenges, with costs exceeding $2.5 billion to bring a single drug to market and success rates hovering between 10-20 percent {source}. This challenging landscape has been further complicated by the industry’s shift from small molecules to complex biologics and cell and gene therapies targeting rare diseases, putting unprecedented pressure on pharmaceutical companies’ internal rates of return (IRR) {source}.
In response, pharmaceutical executives are adopting a technology-forward mindset, focusing on operational efficiency and de-risked investments.
AI-powered precision measures
Artificial Intelligence (AI) powered digital measures represent a promising solution to these challenges such as prioritising pipelines, segmenting disease states, optimising endpoints and composite biomarkers, and selecting precise and appropriate participants for clinical trials. These measures, based on physiological or behavioral sensor data collected from digital health technologies (DHTs) such as wearables and sensors, are transforming how we approach clinical research. The FDA reports reviewing hundreds of AI/ML technology submissions since 2022, documented over 70 case studies that demonstrate their diverse applications in R&D {Source}. Unlocking the power of “deep” personalised medicine, the abundance of these rich and diverse healthcare-data sources can aid in developing a more evidence-backed strategy {source}.
Real-World Impact
The potential of AI-powered digital measures is already being realised. At the 2023 JPM Healthcare Conference, Roche presented compelling evidence of their success with novel digital endpoints. Their implementation resulted in clinical trials for Parkinson’s disease running twice as fast (12 months when using a digital endpoint as compared to 28 months) while reducing sample sizes by 70 percent – from 550 to 166 participants. This efficiency gain, both due to more precise patient selection and shorter trial duration, demonstrates the powerful impact of precision measures on both operational costs and trial execution. {source}.
AI-powered digital measures are also enabling unprecedented insights into disease progression. For instance, researchers have successfully used nocturnal breathing patterns collected at patients’ homes to track Parkinson’s disease progression, demonstrating how these technologies can serve as proxies for clinical progression where traditional biomarkers fall short {source}.
The adoption of these technologies extends beyond individual success stories. A Harvard team analysing ClinicalTrials.gov data found a 34 percent compound annual growth rate in connected digital products use in clinical trials, even before COVID-19 {source}. ICON’s Atlas platform for digital measures has tracked over 3,300 clinical trials utilising DHTs since 2003, involving more than 900 organisations, including 150+ pharmaceutical companies.
Effective Implementation of AI-Powered Digital Measures
Rather than an overnight disruption, the adoption of digital measures has been a steady climb. As with any new technology in a highly-regulated industry such as pharma, rigorous validation is crucial to ensure digital health tools are fit-for-purpose for clinical populations. Generating this evidence to support a DHT’s deployment can be expensive and time-consuming, especially in rare disease populations that are shaping pipelines today and have unique limitations on recruitment {Source}.
To ensure best practices, implementing AI-powered digital measures requires careful consideration and validation. Feasibility, user acceptance, and validation evidence are all necessary to paint the bigger picture of DHT maturity to key stakeholders and effectively implement a precision measures-first approach. For validation, researchers must conduct analytical validation, demonstrating the technology’s accuracy in measuring specific parameters, such as how precisely a sensor can measure gait speed compared to an established reference standard. Equally important is clinical validation, where researchers evaluate the clinical relevance and utility of the data. In this stage, they assess whether the measured parameter serves a meaningful clinical purpose – for instance, in one study they found a correlation between gait cadence from a fitness tracking wearable and health-related quality of life (source).
Leveraging the existing research on the implementation of digital measures and biomarkers can help teams de-risk selection and deployment within their own studies. Resources supporting this implementation journey continue to expand. Pre-competitive consortiums like c-path, patient advocacy groups like Mobilise-D, and FDA guidelines for DHT use provide frameworks for success. Additionally, tools such as ICON’s Atlas platform, which houses over 15,000 pieces of evidence related to DHT maturity, enables researchers to accelerate implementation by leveraging existing validation data.
Looking Forward
AI-powered digital measures represent more than just a new data collection method – they offer a pathway to transform clinical research through improved decision-making, precise patient selection, and flexible trial designs. The evidence suggests that AI-powered digital measures can significantly impact R&D efficiency and effectiveness {source}. Our ability to now leverage artificial intelligence and machine learning to learn from data, identify patterns, and make predictions or decisions is a leap forward for clinical research. As the industry focuses on operational efficiency, executives face a choice: maintain traditional approaches or embrace these emerging technologies to drive precision in clinical research.
For organisations willing to navigate the implementation challenges, these technologies offer a promising solution to the persistent challenges of drug development costs and risks and ultimately, bringing more patient-centred treatments to market.
About the authors
Sonia Bothorel currently serves as the Managing Director of Mapi Research Trust, an industry leader in the distribution and implementation of Clinical Outcome Assessments in global clinical trials. At Mapi Research Trust, Sonia leads a multinational team of 140 employees across operations, sales, marketing, technology and pharmaceutical sciences, and oversees the organisation’s varied revenue streams. A seasoned life sciences executive, she also leads the Outcome Measures practice within ICON, which combines COA data and digital endpoints to enable optimised protocols and patient-centric research. Prior to Mapi Research Trust, Sonia held positions in manufacturing in the pharmaceutical and food industries, including at Nestlé Corporation. She has more than two decades of experience spanning operations, business management, clinical research, and corporate leadership. Sonia holds an Executive MBA from France’s EM Lyon Business School, where she also is a Board Member.
Jess Carruthers is an Outcomes Researcher at ICON, where she works on the Atlas Platform, which offers insight into Digital Health Technologies. Jess holds a Master of Public Health from CUNY School of Public Health.
