Observe Ahead of the Curve

When PicnicHealth Co-founder and CEO Noga Leviner was diagnosed with Crohn’s disease, she was surprised there wasn’t a system that would help manage her condition and share her health data. With the advent of AI, she co-founded her own with the aim of simplifying observational research for life sciences companies.

Observational research is a type of research methodology where investigators observe and record the behaviors, outcomes, or exposures in patients without adding variables or manipulating the environment, as would occur in experimental research. This approach is often used to gather descriptive data while minimizing researcher influence on the subject. But patients are no longer considered subjects in this field. PicnicHealth puts patients in the position of partner, and gives them the tools to help manage their own journeys. Leviner tells us more.

What factors are prompting demand for increased AI capabilities in observational research?
 

Industry inefficiencies, regulatory shifts, and technology advances are driving the need for a new approach to observational research. Too often, traditional, site-based interventional trial structures are a force-fit for observational studies. This is especially important for newer treatment approaches such as cell and gene therapies that require longer term follow-up. Traditional site-based studies put a heavy burden on patients with frequent travel and time commitments. Virtual research makes it easier for patients to participate remotely and contribute to these long-term studies. The FDA recognizes the value of observational research in revealing valuable insights about disease progression and real-world treatment impacts, and is advocating for more patient-centric approaches.

As patients go to more places of care and data lives in more silos, it becomes harder to get full visibility into the patient journey and experience. AI innovations are helping researchers track disease progression and uncover insights much faster than manual approaches. The industry has historically been risk-averse and slow to change, but biopharma companies are quickly embracing AI and advanced technologies to capture patient experience data, and making it central to study design and drug approval processes. This not only improves research, but also reduces a lot of the burden on patients and sites.

What are the major challenges faced by observational researchers, and how can new digital tools help?
 

The main challenge is that observational research remains constrained by the same inflexible processes, tools, and standard operating procedures used in interventional clinical trials. Embracing digital and AI tools can really improve speed and efficiency.

The traditional observational research model fails to reach patients where they are, limiting access to valuable real-world data. Because of this limitation, studies take longer and, in a budget-constrained world, sponsors conduct fewer studies.

Observational studies are a good starting point for accelerating development and leveraging advanced digital tools and AI. Virtual clinics and AI-powered data analysis can finally bridge the gap between the realities of the real world and scientific rigor, overcoming traditional barriers to research.

These innovations provide researchers complete and continuous visibility into patient data and outcomes beyond what is captured in electronic medical records. A more efficient research process also reduces the burden on both patients and sites, making it easier for patients to participate and sites to run more studies.

In what other ways is AI simplifying research for clinical research and the patient journey?
 

AI can advance research by automating time-consuming research tasks, such as patient record screening and study eligibility assessment. These technologies capture and analyze large volumes of clinical data to uncover new patterns, enhance study designs, and meet regulatory standards. This is transforming researchers’ roles in producing high-quality, informative data. AI and deep learning techniques now understand human language and extract clinical observations from medical records, including a clinicians’ unstructured narrative notes. This frees up scientists to focus on more complex aspects of study design and analysis.

In terms of the patient journey, the industry is still in the early stages of fully tapping AI’s potential for understanding patient experiences. Studies are more impactful when they aren’t limited to a single clinic, hospital, or healthcare system, and instead incorporate data and insights from across the patient journey – from risk perceptions and symptom management to disease progression and treatment side effects. Traditional site-based research models struggle to meet these demands of modern evidence generation.

How can the industry ensure AI doesn’t become too intrusive? And how important is it to keep humans in the loop?
 

AI excels at automating and accelerating existing research processes, but its implementation should always be carefully evaluated on a case-by-case basis. For lower-risk, routine tasks, such as initial patient record screening, AI is a better alternative that can free up human resources. However, more complex and sensitive decisions still require human judgment and oversight. When it comes to submitting endpoints for regulatory submissions, for example, experienced researchers are necessary to review and validate AI-generated insights before they are submitted.

As we move forward, the industry must thoughtfully and strategically deploy AI while prioritizing patient privacy through anonymized data and research integrity. Rather than replacing human researchers, AI can augment and enhance the research process. It’s about striking a balance between AI handling high-volume, repetitive tasks while keeping control over critical decisions and oversight in the hands of experienced researchers.

What are the shortcomings in decentralized clinical trials as we know them, and how do you expect a virtual clinic to address them?
 

A virtual clinic won’t solve all the challenges, but it is one important piece of the puzzle. There are clear advantages to moving research outside of the traditional site-based model, including flexibility, lower costs, reduced site and patient burden, and speed, to name a few. But there are real limitations with virtual study designs meeting appropriate scientific, ethical, and technical requirements. A virtual clinic can bridge this gap. It’s one important tool in the toolbox to expand the range of studies where a virtual design is appropriate.

These clinics enable remote patient engagement, data collection, and clinical assessments that are complementary to traditional virtual research tools like patient-reported outcomes (PROs) and wearables. When combined with access to complete medical records across all a patient’s sites of care, virtual clinics capture meaningful data that goes beyond traditional decentralized trials.

Give us a bold prediction of what clinical research could look like ten years from now…
 

Within the next decade the line between research and real-world clinical practice will blur and, ultimately, shift from solely discrete, carefully controlled trials that are extrapolated to the broader population to research that will be much more of a continuous learning process.

Data generated during routine patient care – from electronic health records (EHRs), connected devices, virtual visits, and other sources – will easily feed into research efforts. Researchers will be able to tap into this rich, longitudinal data to gain deep, personalized insights about individual patients beyond the limitations of small, randomized studies. At the same time, the research process itself will be far more integrated into patients’ everyday lives. Virtual clinics, remote monitoring, and other digital tools will make it dramatically easier for patients to participate in studies without the burden of frequent site visits. The patient experience will be a core focus, not an afterthought.

AI and advanced analytics will play a central role, automating many of the repetitive, high-volume tasks in research, while researchers remain essential to maintaining human oversight and judgment. AI will augment and enhance the work of researchers, not replace them entirely.