If you are looking for a step-by-step discussion on some of the data pain points of clinical trials, then the episode of Data in Biotech with Miguel Cacho Soblechero is a great starting point. Looking at every stage, we sat down to discuss the challenges of clinical trials and the direction data science is steering the biotech and pharma industries.
Miguel comes to data science from the world of engineering. Beginning his career programming medical devices, Miguel became passionate about the intersection of engineering, data, and healthcare and how these disciplines can work together to improve lives. This led him back into academia, completing a Master’s and PhD at Imperial College, London, before moving to Boston for his current role at Prolaio, a HealthTech organization that uses data to drive better outcomes for patients with heart conditions.
Alongside an overview of how a data team can support each step of a clinical trial, we chatted to Miguel on how his career to date influences his current work, how data science can level the playing field for smaller companies, and big industry trends like AI. Here are the highlights:
- The Human Factor (2:15): Miguel talks about the transition from academia into industry and how the big shift in focus required is understanding the people who will use the models and analyses you produce. You are no longer focused purely on developing the best model, achieving the highest accuracy, and getting the most from the data. You also have to consider how the data is going to be used to impact the life of the end user. This aspect must come first, and once you understand the outcome you are trying to drive from a human perspective, then you can go about developing something to serve this goal.
- Broadening the pool for clinical trials (10:03): Clinical trials, Miguel explains, can be limited by a team’s knowledge of eligible patients for the trial. In many cases, they rely on personal networks, which limits the pool of people we can draw from. However, by using anonymized, aggregated data using Electronic Health Records (EHR), a data science team can help their clinical team access a much broader range of patients that would have been previously unknown. This not only benefits patients, but adds diversity to the trial.
- Tackling dropout rates (12:26): There is a huge problem for clinicians running clinical trials with patient dropouts. Miguel suggests that similar techniques to those used by retailers to encourage purchases can be used by clinical trials to drive continued engagement. Data can be used to predict which patients are likely to “churn” (cease participation) and intervene to ensure that the data they are in the process of generating can be utilized in the final results of the trial. Likewise, if a participating clinic or patient is lagging behind on submitting data and the trial organizer is aware, they can step in, understand why, and solve the problem with the clinic or the patient, before it is too late. This fosters a proactive rather than reactive culture and goes back to understanding and accounting for the human factor.
- Reducing risk with data (27:40): Miguel states that clinical trials are possibly the biggest risk factor for any pharma company. However, if we can use AI to mitigate the risks that would cause a trial to be unsuccessful, we can reduce the failure rate. He goes on to explain the implications of this across the entire chain, “We've been only talking about clinical trials, but we can talk about drug discovery. If we're able to create better drugs and understand those drugs better from the inception. If we can reduce the costs of running a clinical trial and the time that it takes. That's going to have a knock-on effect on price and on risk.” This is of real benefit to smaller companies and has the potential to significantly level the playing field.
- Inclusion of wearables to improve clinical trials (31:38): When asked about the future of clinical trials, Miguel talks about the value of wearables. He explains if you have to go to the hospital three times a week for a clinical trial, it would be time-consuming and a major factor in patient churn. However, if you can be monitored at home, that will massively increase patient compliance. It will also provide much richer, time-series data to build a more complete picture of how the patient is reacting to treatment during a trial.
Further reading: Statistics Done Wrong by Alex Reinhart was Miguel’s recommendation at the end of the episode as a light read that is quite funny and really insightful.
Continuing the Conversation
One of the aspects that we touched on with Miguel was the limitations of EHR, and the importance of a proprietary dataset that is internally curated and internally generated. He commented on the advantage big pharma has in being able to build this type of dataset.
Although we looked a little at what smaller biotechs can do to level the playing field when it comes to how they use their data, let's explore this in more detail, specifically the topic of collecting data. How can smaller pharma and biotech organizations working on clinical trials make sure they are building valuable internal data sets that can support their business goals?
- Identify the data you want to capture across clinical trials: Start by identifying the specific types of data that will be valuable to your organization. This data may include patient demographics, clinical outcomes, safety (i.e. side effects and toxicity), laboratory results, and other relevant variables. Ensure that these data elements align with your research goals and broader business objectives. For example, capturing patient-reported outcomes (PROs) can provide valuable insights into the patient experience and treatment efficacy.
- Build consistent data collection infrastructure and processes into your clinical trial protocol design: When designing clinical trial protocols, embed consistent data collection infrastructure and processes directly into each protocol. Specify consistent types of data to be collected, timing of data collection, and methods for data capture. This proactive approach ensures that data collection is systematic and consistent across all trials.
- Develop automated, canned analyses for regular review: Create automated, canned analyses or dashboards that the clinical operations team can review regularly during trial implementation. These analyses should include key performance indicators (KPIs), safety signals, and efficacy assessments. Schedule regular monitoring meetings and create thresholds and alerts for each indicators to trigger ad-hoc review. Regular monitoring of these predefined metrics can help identify issues early, allowing for prompt intervention. Examples of analyses could include patient recruitment rates, adverse event monitoring, and protocol adherence metrics.
If you're interested in discovering how your organization can unlock the value of data and maximize its potential, get in touch with CorrDyn for a free SWOT analysis.
Want to listen to the full podcast? Listen here: