Podcast Summary
Exploring the Future of Digital Diagnostics: Digital diagnostics use AI and other technologies to analyze medical data for accurate and efficient diagnosis. Benefits include improved patient outcomes, cost savings, and increased efficiency. Incentives include regulatory approvals and market demand. Payment models for adoption are evolving.
Digital diagnostics refer to advanced diagnostic tools that leverage artificial intelligence (AI) and other technologies to analyze medical data, including clinical information, digital pathology images, and genomic data, for more accurate and efficient diagnosis. The discussion on Clubhouse, led by Vinny Tagrawala from Andreessen Horowitz, involved Gaurav Singhal, a physician and computer scientist with expertise in this area. The goal was to explore the future of digital diagnostics and its implications. The conversation touched upon three main aspects: defining digital diagnostics, understanding the benefits and incentives, and discussing the payment models for their adoption. Gaurav, who has experience in creating NLP solutions and overseeing a large data platform at Foundation Medicine, provided valuable insights into the potential of digital diagnostics in revolutionizing healthcare. The conversation was recorded and open for audience participation, with the aim to bring in additional speakers and answer questions.
Digital diagnostics: Reinterpreting healthcare data for new insights: Digital diagnostics use existing healthcare data to generate diagnostic, predictive, or prognostic insights, with potential regulatory differences from traditional tests. Blurring lines between traditional and digital diagnostics, cost savings, and regulatory considerations make them a game-changer in healthcare.
Digital diagnostics, also known as second order diagnostics, refer to the reinterpretation of existing healthcare data to generate new insights, which can be diagnostic, predictive, or prognostic. These insights can come from various data types, including path image data, EKG data, telemetry data, radiology data, and ICU data. The regulatory path for digital diagnostics can vary, with some following the same path as traditional laboratory developed tests, while others may not be regulated in the same way. The boundaries between traditional diagnostics and digital diagnostics are becoming increasingly blurred, as many diagnostic tests, including radiology and pathology, are digital by nature. Reinterpreting existing data can also have practical implications, such as reduced costs and a shift in challenges to accessing and interpreting the data. The regulatory landscape for digital diagnostics is still evolving, and the FDA is considering various dimensions that make these diagnostics unique. Overall, digital diagnostics represent a new wave of emerging diagnostics and insights that have the potential to revolutionize healthcare.
A shift towards digital diagnostics: The diagnostic industry is transitioning from acquiring new samples to accessing and utilizing existing data, expanding the ecosystem and leading to real clinical relevance through machine learning and computational methods
We are witnessing a shift in the diagnostic industry towards a more digital age. Instead of focusing on obtaining new samples, the emphasis is now on accessing and utilizing existing data. This change expands the ecosystem, making various entities like information systems and diagnostic industry players crucial participants. While accessing large training datasets may require significant effort in some cases, it can be less complex than developing new experimental diagnostic tests. The availability of digitized data and technological advances in machine learning and computational methods are leading to real clinical relevance and capabilities of digital diagnostics. However, the realization of this potential in clinical care has yet to be fully achieved. Despite seeing some progress, we're not there yet, but the golden age of digital diagnostics is on the horizon.
Transitioning diagnostics to a digital landscape: Advancements in sensor tech and emergence of home-based diagnostics generate high-quality data, aiding in diagnosis for rare conditions and expanding pharma patient bases.
The transition of diagnostics to a more digital landscape faces challenges, including regulatory hurdles and the difficulty of generating high-quality, utility-proven datasets. However, this shift could bring significant benefits, especially for the pharmaceutical industry. With the advancement of sensor technology and the emergence of home-based and smartphone diagnostics, high-quality data is now being generated in areas such as EKGs and digital pathology. This data can aid in diagnosis, particularly for rare conditions where effective therapies are emerging but diagnosis remains challenging. Pharma companies stand to benefit greatly by gaining access to a larger patient base for their medicines. The adoption of digital diagnostics may be slower than other areas of healthcare due to regulatory and incentive structures, but the potential benefits for both patients and industries make it an exciting area to watch.
Identifying undiagnosed patients using second-order diagnostics: Pharmaceutical companies can leverage existing data for diagnostics, benefiting patients, providers, institutions, payers, and industry. However, payers should determine diagnostic validity and cost-effectiveness to prevent unilateral drug sales.
There is a significant opportunity for the pharmaceutical industry to identify and treat undiagnosed patients using second-order diagnostics derived from existing data. This could lead to substantial benefits for patients, healthcare providers, institutions, payers, and the pharmaceutical industry itself. However, there are concerns that if pharmaceutical companies become the gatekeepers of the diagnostics industry, it could lead to a unilateral adoption of diagnostics with the sole purpose of driving drug sales. Instead, it is suggested that a traditional payer system and regulatory framework be used to determine the soundness and justifiability of diagnostic tests against their costs. Payers could identify patients for earlier and more accurate diagnoses, reducing long-term care costs, and preventing unnecessary therapies. Ultimately, the goal should be to prioritize patient care and well-being over commercial interests.
Trend of radiomic signatures as second order diagnostics for checkpoint inhibitors: Pharma companies can drive adoption of radiomic diagnostics for checkpoint inhibitors by investing in their development and implementation, ultimately creating value for the entire ecosystem.
There is a growing trend in the oncology space for companies developing radiomic signatures as second order diagnostics for checkpoint inhibitors. These diagnostics, based on the appearance of tumors on CT scans, can predict which patients are likely to respond or not, potentially saving payers from unnecessary spend on ineffective treatments. However, the implementation of these diagnostics may depend on whether they are seen as companion diagnostics requiring wet lab testing or real-world evidence analytics. While payers may eventually benefit from these diagnostics, the speaker argues that pharma companies need to take the initial risk and drive adoption to make these benefits a reality. The speaker suggests that pharma companies can subsidize the adoption of these diagnostics by investing in their development and implementation, ultimately creating value for the entire ecosystem.
Investing in diagnostic infrastructure for cost savings and benefits: Pharma companies can invest in diagnostic infrastructure for cost savings and benefits, including creating core algorithms and providing data collection infrastructure. This can help identify patient populations and potentially recoup investment.
The development, distribution, and reimbursement costs are crucial factors in bringing a novel diagnostic to market. While traditional diagnostics and digital diagnostics share some common cost categories, there are unique costs associated with digital diagnostics. Gaurav suggests that pharma companies should invest in the development of foundational diagnostic infrastructure, which can lead to cost savings and benefits for both parties. This investment could include creating the core diagnostic algorithms and providing the necessary infrastructure for data collection and analysis. By doing so, pharma companies can identify patient populations for their drugs and trials, and potentially recoup their investment through the value generated downstream. However, it's important to note that the development of this infrastructure is just one piece of the puzzle, and further research is needed to determine who should be responsible for the platform-wide development costs.
Integration of Pharma and Diag Companies Advancing Healthcare: The integration of pharmaceutical and diagnostic companies holds great potential for advancing healthcare through the development of diagnostic platforms, leading to the creation of various diagnostic tools and subsidization of initial diagnostics.
The integration of pharmaceutical companies and diagnostic companies holds great potential for advancing healthcare through the development of diagnostic platforms. These platforms can lead to the creation of various diagnostic tools, including those for specific conditions like hypertrophic cardiomyopathy, cardiac amyloidosis, and malignant arrhythmias. Pharma's involvement in subsidizing these platforms and initial diagnostics is a no-brainer due to the significant value generated on the therapeutic side. However, the process of integration is complex, and there are different approaches, such as vertical integration or diagnostic companies seeking a share of therapeutic profits. The success of these strategies remains to be seen, and the jury is still out. Additionally, companion diagnostics, which are diagnostics that provide access to specific therapies, may not be the only application for these platforms, as they can also produce diagnostics unrelated to specific therapies in the future. Overall, the integration of pharmaceutical and diagnostic companies can lead to significant advancements in healthcare, but the path forward is complex and uncertain.
Value-based care system and diagnostic industry evolution: The diagnostic industry's role in value-based care is crucial, but equitable distribution of value between diagnostics and pharma companies is necessary to save costs and improve patient outcomes. Solutions include value-based pricing, a value-based ecosystem, and addressing attribution complexities.
The healthcare industry is in the process of transitioning towards a value-based care system, where preventative diagnostics that identify diseases earlier can save the healthcare system money overall. However, there is a missing piece in this ecosystem: an equitable distribution of value between diagnostic companies and pharma companies. This issue not only affects fairness but also presents an opportunity cost. If we can find a solution to this, the industry could accelerate and benefit the world. This solution could involve value-based pricing for pharma assets and a value-based ecosystem where all parties share the cost savings. However, there are scenarios where diagnostics may not save costs but instead drive increased utilization, leading to more expenses for insurers. Another challenge is attribution, which is crucial for determining the value of diagnostics and distributing it fairly among all parties involved. The industry is evolving rapidly, and the opportunity for partnerships between providers, payers, and diagnostic companies is unprecedented. The FDA's willingness to regulate medical software as a device and consider adaptive software could further decouple diagnostics from pharma. It's essential to consider these complexities as the industry moves towards a more value-based care system.
The impact of diagnostic tools on patient outcomes: Diagnostic tools save costs and improve patient outcomes by identifying eligible patients for life-saving procedures and incentivizing providers through reimbursement. However, the use of adaptive algorithms in digital diagnostics raises complex safety and regulatory concerns.
The role of diagnostic tools and tests in healthcare extends beyond cost savings and can lead to significant improvements in patient outcomes. The example of Viz.ai's stroke diagnostic tool illustrates this, as it identifies patients eligible for life-saving procedures and incentivizes providers through reimbursement. However, the use of adaptive algorithms in digital diagnostics raises complex questions around safety and regulation. These algorithms constantly learn from new data, making traditional deterministic evaluation frameworks insufficient. The challenge lies in creating a robust infrastructure to assess and ensure the safety and efficacy of these adaptive algorithms. It's an exciting yet daunting prospect that requires careful consideration and collaboration among stakeholders in the healthcare industry.
Regulatory challenges in adopting imaging analysis algorithms in medicine: Despite regulatory hurdles, consensus is that early versioning and testing of AI updates in medicine is acceptable, with a focus on solving other problems first, and balancing human expertise and AI capabilities in clinical decision-making.
The adoption of imaging analysis algorithms in medicine is facing challenges beyond the complexity of the algorithms themselves. The regulatory burden, specifically the FDA's guidance on software as a medical device, is a significant hurdle. However, the progressive nature of this guidance shows promise for more efficient updates in the future. Despite the regulatory challenges, there is a consensus that the worst-case scenario of versioning and testing each update is acceptable in the early stages of AI implementation in medicine. Companies like Foundation Medicine have already implemented this approach. The focus should be on solving other problems before tackling online updating of algorithms. While the potential for algorithms to improve and learn is exciting, there are concerns about their ability to fail spectacularly and the serious implications that could have on patients' lives. The FDA's consideration of this issue is appreciated, but it's seen as a future problem rather than a pressing one. The clinical decision-making process today relies on clinicians' experiences and learning from each case, and getting rid of them altogether is not the solution. Instead, the goal should be to find a balance between the human expertise and the capabilities of AI algorithms.
Acknowledging the limitations of current diagnostic methods: It's essential to continue the conversation around complex diagnostic challenges, acknowledging limitations while striving for improvements.
There's an unrealistic expectation for computational and genetic diagnostics to be flawless and controlled, while the current methods are inherently messy. This was compared to driverless cars, which are expected to never crash. The standard for non-human systems is set much higher. Despite the challenges, it's essential to continue the conversation around these complex issues. A big thank you to everyone who participated in the discussion, especially Gaurav for joining. Although I stumbled at the beginning of the Clubhouse event, it ultimately proved to be an engaging and insightful experience. The conversation highlighted the importance of acknowledging the limitations of current diagnostic methods while striving for improvements. Thank you all for your thoughtful questions and participation. Goodnight!
