You have 4 summaries left

The Gradient: Perspectives on AI

Shiv Rao: Enabling Better Patient Care with AI

Thu Jul 13 2023
AI in HealthcareGenerative AIClinical DocumentationWorkflow OptimizationHealthcare Technology


This episode explores the intersection of AI and healthcare, focusing on the potential of generative AI technology in improving patient experiences and clinical encounters. The guest, Shiv Rao, a practicing cardiologist and founder of Abridge, shares insights on the challenges and opportunities in healthcare, the importance of trust and transparency, and the role of technology in transforming workflows. The episode highlights the impact of Abridge's real-time note-taking technology in saving time for doctors, improving accuracy of clinical documentation, and enhancing doctor-patient relationships. It also discusses partnerships, data collection, reinforcement learning, and the future of AI in healthcare.


AI's Potential in Healthcare

AI has the potential to automate inefficiencies in healthcare, improve patient experiences, and address clinician burnout and labor shortages.

Abridge Technology

Abridge is a generative AI technology that finds the signal in healthcare conversations, generates artifacts in real time, and improves doctor-patient interactions.

Partnerships and Trust

Abridge is partnering with institutions to create a platform for clinical trials, building trust through randomized control trials, and positioning its product as an assistant rather than a replacement for doctors.

Workflow Optimization

Transparency, data partnerships, and careful consideration of technologies are crucial for workflow optimization in healthcare.

Technology Adoption

When adopting technology in healthcare, it's important to consider leveraging existing tools, incorporating objective feedback, and aiming for seamless integration into existing workflows.

AI in Healthcare

AI has the potential to create value, improve outcomes, address workforce shortages, and enable healthcare workers to focus on clinical decisions.

Multidisciplinary Collaboration

Collaboration between healthcare experts and AI experts can lead to valuable projects at the intersection of AI and healthcare.

The Future of Healthcare

Technology will play a central role in shaping the future of healthcare, and open-mindedness towards new technologies and solutions is crucial.


  1. Introduction
  2. Tailwinds in Healthcare
  3. Abridge Technology
  4. Partnerships and Trust in Healthcare
  5. Transparency and Workflow Optimization
  6. Data Collection and Reinforcement Learning
  7. Technology Adoption in Healthcare
  8. AI in Healthcare
  9. Conclusion


00:00 - 08:08

  • Medicine is an exciting field where new technology can make an impact, but turning imagination into reality is challenging.
  • Predictions of AI replacing radiologists and making clinical work more efficient have not yet come true.
  • Shiv Rao, a practicing cardiologist, understands the healthcare issues and the potential of AI in medicine.
  • Abridge, a company led by Shiv Rao, has partnered with the University of Kansas to bring generative AI to the healthcare space.
  • Shiv Rao's background includes being a corporate VC for a large hospital system and investing in technology aimed at helping doctors and nurses.
  • He also invested in Carnegie Mellon University's machine learning and health program, which influenced the founding ideas of Abridge.
  • Shiv Rao's interest in medicine started when he was inspired by an ophthalmologist who helped millions of people regain eyesight through innovative methods.
  • During his cardiology fellowship, Shiv became passionate about combining healthcare and technology, particularly machine learning and AI.
  • AI has the potential to automate inefficiencies in healthcare and improve patient experiences.
  • Healthcare has multiple stakeholders with misaligned incentives, making it difficult to create transformative solutions. Abridge aims to put the patient experience at the center of its solutions.
  • Regulatory mandates have been valuable tailwinds for companies that have made significant impacts in healthcare.

Tailwinds in Healthcare

07:43 - 15:14

  • The companies who have inflected in healthcare can be traced back to regulatory tailwinds or historic moments intersecting with technology.
  • Telemedicine during the pandemic allowed some companies to quickly shift to virtual care and capitalize on the opportunity.
  • There is a tailwind around clinician burnout and labor shortages, with an estimated 63% of clinicians experiencing burnout.
  • Hospital systems can't buy themselves out of the crisis and need deflationary tools that can scale and democratize value proposition.
  • The gender of AI is another tailwind in healthcare, as health systems recognize the need for productivity tools that improve staff satisfaction.
  • Technology in healthcare can make it feel more human again by allowing more time for personal relationships and idealistic practices like home visits.
  • Abridge technology enables real-time note-taking during patient visits, improving rapport and reducing administrative burden on doctors and nurses.
  • Healthcare has been plagued by garbage-in-garbage-out machine learning due to rushed data entry processes by clinicians.
  • Abridge technology changes the game by taking notes in real time, allowing for more meaningful conversations between doctors and patients.

Abridge Technology

14:45 - 22:10

  • Abridge is a technology that finds the signal in healthcare conversations and removes small talk
  • It generates artifacts for different constituents in healthcare, such as providers, care team members, payers, and patients
  • Using generative AI, Abridge creates different artifacts in real time to improve patient conversations
  • Doctors have more time to focus on patients and build better relationships
  • The technology improves the accuracy of clinical documentation and frees up time for doctors
  • A doctor at University of Kansas compared Abridge to the stethoscope in terms of significance
  • Abridge enables doctors to be more present and focused on their patients
  • The goal is to improve clinical encounters by enabling people to operate at the top of their license
  • University of Kansas has over 1500 clinicians using Abridge and saving up to two hours a day on documentation
  • Abridge can scale across different specialties and benefit all types of clinicians
  • The company's academic DNA allows for rigorous research and development in healthcare
  • They are focused on demonstrating the impact through trials and building trust

Partnerships and Trust in Healthcare

21:52 - 29:07

  • The company is focused on building trust in new medications and techniques through randomized control trials.
  • They are partnering with various institutions, including the University of Kansas and UPMC, to create a platform for clinical trials.
  • They will soon announce a data partner who will provide them with previously unseen annotations on workflows.
  • Trust, reliability, transparency, and credibility are crucial in healthcare.
  • The company positions its product as an assistant and augmenter rather than a replacement for doctors.
  • There is still a lot of art and decision-making involved in medicine that cannot be fully automated.
  • The company recognizes the efficiencies that can be created through automation in low-risk, high-volume decisions.
  • Humility is important when developing health tech startups to avoid fatal flaws.
  • There is an opportunity for AI in healthcare to create incredible efficiencies in certain workflows.

Transparency and Workflow Optimization

28:59 - 36:25

  • Transparency is an important aspect for healthcare startups like Bridge
  • Bridge's user interface allows clinicians to map every generated summary and identify parts of the conversation that were not covered
  • Bridge continuously learns from user annotations and edits to improve its technology
  • Partnerships with companies providing higher-level data sets help Bridge stay ahead in healthcare
  • In the future, every company will likely be an AI company to some extent
  • There are two categories of companies: those leveraging APIs and those building their own AI models
  • Building homegrown models is valuable in healthcare due to the industry's barriers to entry
  • The perspective of solving problems through algorithmic solutionism may differ between computer scientists and companies like Bridge

Data Collection and Reinforcement Learning

35:56 - 43:53

  • The company believes in bringing together doctors, machine learning engineers, designers, and product people to work towards delivering better patient outcomes.
  • They have aggregated a large unstructured data set of medical conversations to benchmark their performance and put in front of clients and partners.
  • The company takes a longer time horizon when it comes to AI development and has both vertical solution and horizontal R&D aspects.
  • Generative AI technologies are quickly moving up the stack in terms of value, but guardrails need to be created for specific use cases.
  • Detailed datasets are crucial for various healthcare applications, and an AI native company needs to collect, clean, aggregate, and learn from them.
  • Reinforcement learning can be useful in healthcare documentation to ensure notes sound trustworthy and align with the provider's style.
  • Careful consideration is needed when applying different technologies in different areas of healthcare.

Technology Adoption in Healthcare

43:29 - 50:44

  • When applying different technologies in healthcare, it's important to consider whether to build your own stack or leverage existing tools.
  • Feedback that is objective and aligns with specific ontologies in healthcare can be safely incorporated into generative AI models.
  • Qualitative and stylistic feedback requires caution to avoid steering the model in a direction that doesn't align with healthcare goals.
  • In the space of clinical documentation, there are opportunities for innovations and improvements beyond cost and time savings.
  • Workflow optimization plays a crucial role in healthcare delivery, from assigning medical record numbers to generating images and making treatment decisions.
  • Unbundling care delivery modules and creating standalone solutions is not sufficient to transform healthcare due to barriers to entry and misaligned incentives.
  • Generative AI solutions aim to seamlessly integrate into existing workflows, improving efficiency without disrupting user experience.
  • The goal is for users to perceive the innovation as invisible, where their workflow remains unchanged but operates more effectively.

AI in Healthcare

50:16 - 58:10

  • Capital D design point of view is important in creating value and justifying workflow adjustments in healthcare.
  • The goal is to have more human interactions and less interactions with technology in healthcare.
  • Some AI researchers make wild predictions about the applications of AI in healthcare, but the realities on the ground are different.
  • There is a lot of opportunity to augment and automate in healthcare, which can improve outcomes for everyone involved.
  • Leveraging technology can help address the shortage of skilled workers in healthcare.
  • Automation should enable healthcare workers to operate at the top of their license and focus on clinical decisions rather than clerical work.
  • The intersection of AI and healthcare offers promising ways to create value and change the game.
  • People interested in exploring this intersection should absorb information about AI and healthcare, connect dots across disciplines, and collaborate with experts from both fields.


57:42 - 1:00:50

  • The combination of healthcare experts and AI experts can lead to valuable multidisciplinary projects.
  • Healthcare systems are now open to trying new technologies and solutions.
  • This is an opportunity for the best products, technologies, and user experiences to succeed.
  • Technology will play a central role in shaping the future of healthcare.
  • The guest's perspectives provide a grounded and valuable perspective on what's feasible and happening in the industry.
  • The podcast episode concludes with information about subscribing to The Gradient and leaving feedback.