Drug development in 2025: 5 expert predictions cover synthetic data, hybrid trials and more 

Jay Anders, MD, chief medical officer of Medicomp Systems, which makes medical data relevant, usable and actionable, foresees a seismic shift in how the industry approaches AI model training.

“In 2025, we’ll see a significant pullback in the use of synthetic data for AI model training in drug development, as the industry recognizes the limitations and potential risks highlighted by recent FDA-approved medical devices. Drug developers will increasingly prioritize high-quality, real-world patient data for AI training, leading to more reliable and clinically validated drug discovery processes.“

The integration of AI in clinical trials will reach new heights, according to Kim Perry, Chief Growth Officer at emtelligent.

“In 2025, we’ll see AI drive transformative changes in drug discovery, with synthetic data playing a major role in refining trial design and early-stage analysis, accelerating timelines, and enabling precision-driven protocols. Expect that more than half of new trials will incorporate AI-driven protocol optimization to address long-standing recruitment and engagement hurdles with an efficiency the industry has barely glimpsed so far. Predictive analytics will elevate site selection and patient recruitment, following precedents set in oncology, where AI-driven insights from genomic and clinical data have enabled precise cohort identification and real-time treatment adjustments. Regulatory frameworks will likely evolve to consider the use of synthetic data to augment real-world evidence in trials, setting new benchmarks for trial consistency and transparency. As pharma deepens collaborations with AI innovators, technologies like federated learning will enable secure, multi-site data sharing, protecting patient privacy while generating insights at scale. Advanced AI will soon be indispensable, integrating vast EHR and medical device data to reveal new correlations—from individual genomics to population health—bringing us closer to truly personalized medicine.“

Brent Dover, CEO of Carta Healthcare, sees 2025 as a turning point for AI-powered clinical trials.

“I believe 2025 will be the year drug researchers begin harnessing the power of AI at scale to accelerate clinical trials and get life-saving treatments into the hands of patients faster. Quality data is the most important ingredient for improving healthcare, but that data often is buried in unstructured clinical notes. Fortunately, AI-based tools are available today that, along with the help of clinical experts ‘in the loop,’ enable providers to leverage insights from clinical data for drug trials and medical registries. Expect more providers to adopt these clinical data abstraction tools in the coming year and beyond.“

Michael Young, Co-Founder of Lindus Health, predicts a major shift toward hybrid trial models.

“Tools like natural language processing and predictive analytics will allow us to reach and engage people in a way that feels more personal and relevant to them. I see hybrid trials becoming the norm, especially in chronic disease, where decentralized models are a natural fit. Real-world data will also play a huge role in reshaping trial design, letting us adapt in real time based on how patients are actually doing in their day-to-day lives.“

The field of psychiatric drug development will see significant advances in biomarker validation, according to Greg Hajcak, PhD, Chief Scientific Officer of Universal Brain and Sheri Sobrato Professor at Santa Clara University.

“We will see more significant activity and consensus in 2025 around the types of data and interpretable biomarkers studied to support the development of psychiatric drugs. Pharmaceutical and life sciences companies have been exploring different types of biomarkers in recent years – memory, cognitive performance, genes, voice, and many others – to serve as scientifically valid data that can be tested and retested in clinical trial settings. Among the most promising biomarkers are event-related potentials, which are functional brain measures proven in numerous peer-reviewed studies to be highly reliable, consistent and interpretable. We can expect to see a much broader application of these biomarkers before, during and after clinical trials in the new year and beyond.“