Scribbler Logo

    You have 4 summaries left

    This Week in Startups

    Next Unicorns: Bringing AI-powered solutions to medical analysis with PathAI CEO Andy Beck | E1780

    Wed Jul 19 2023
    PathologyDiagnosticsDigitizationConnectivityAICancer DiagnosisTreatment EfficacyCancer DetectionPreventionData-Driven Healthcare

    Description

    The episode discusses the transformation of pathology and diagnostics through advances in digitization and connectivity. It explores the challenges in implementing technology, the use of AI-based measurement algorithms for cancer diagnosis, and the improvement of treatment efficacy with AI. The episode also delves into advancements in cancer detection and prevention, data-driven projects in healthcare, and the future of pathology and diagnostics. The speaker's unique perspective in connecting industries and companies is highlighted.

    Insights

    Advances in Pathology and Diagnostics

    Advances in digitization and connectivity are transforming the field of pathology and diagnostics, improving speed and accuracy in cancer diagnosis and treatment.

    Challenges in Implementing Technology

    The resistance to new technology in the medical field comes from its newness and the challenge of moving and analyzing large amounts of images.

    AI-Based Measurement Algorithm for Cancer Diagnosis

    AI can provide immediate results and accurately classify cells based on various criteria, improving accuracy, reproducibility, and efficiency in pathology.

    Improving Efficacy of Treatments with AI

    AI can analyze quantitative features before and after treatment, helping assess the effectiveness of new drugs for diseases like NASH.

    Advancements in Cancer Detection and Prevention

    Advancements in early detection and risk factor prevention are making cancer more manageable, but screening can lead to unnecessary procedures and anxiety.

    Data-Driven Projects in Healthcare

    Pooling data from various sources and taking diverse approaches can lead to significant advancements in healthcare.

    The Future of Pathology and Diagnostics

    Cloud and digital technology are transforming the field, democratizing access to expertise and improving accuracy, reproducibility, and predictability.

    The Speaker's Unique Perspective

    The speaker's ability to connect with entrepreneurs and investors, drawing connections between industries, provides a unique perspective.

    Chapters

    1. Advances in Pathology and Diagnostics
    2. Challenges in Implementing Technology
    3. AI-Based Measurement Algorithm for Cancer Diagnosis
    4. Improving Efficacy of Treatments with AI
    5. Advancements in Cancer Detection and Prevention
    6. Data-Driven Projects in Healthcare
    7. The Future of Pathology and Diagnostics
    8. The Speaker's Unique Perspective
    Summary
    Transcript

    Advances in Pathology and Diagnostics

    00:00 - 07:31

    • The field of pathology and diagnostics for drug development and clinical care will be transformed by advances in digitization and connecting brains across the world.
    • Cloud and digital connectivity are huge platform advances that will facilitate this transformation.
    • PathAI is a research platform designed to improve the speed and accuracy of cancer diagnosis and treatment using AI.
    • Deep learning has made a significant impact in the interpretation of medical images, which play a crucial role in medicine.
    • Pathologists examine tissue samples under a microscope to make image-based diagnoses of diseases like cancer.
    • PathAI analyzes large whole-slide images of tissue biopsies to accurately diagnose diseases like cancer and predict disease aggressiveness and treatment response.
    • Many labs still use microscopes for generating images, but there is an increasing shift towards digital imaging systems that can capture images digitally and send them for analysis in the cloud.
    • Currently, about 90% of pathology work is done offline, but this is gradually changing as more labs adopt digital solutions.

    Challenges in Implementing Technology

    07:07 - 13:41

    • It depends by country, but about 90% of healthcare is done offline and only about 10% is digital.
    • There is a need for companies to work on capturing more digital data in areas like pathology.
    • Pathology images are data-rich and can be used for deep learning models.
    • A company has built a community of pathologists to train models using expert annotations.
    • The resistance in the medical field comes from the newness of technology and the challenge of moving and analyzing large amounts of images.
    • Technology and deep learning image analysis are new in the field of pathology.
    • The challenge lies in implementing the technology and making it accessible to pathologists.
    • Using algorithms, images can be distributed instantly, unlike physical glass slides.

    AI-Based Measurement Algorithm for Cancer Diagnosis

    13:25 - 20:33

    • There is a profit incentive for affluent individuals to seek second opinions using this technology.
    • The demo showcases an AI-based measurement algorithm for PD-L1 protein expression in lung cancer.
    • Pathologists currently manually estimate protein expression, which is a difficult task.
    • AI can provide immediate results and accurately classify cells based on various criteria.
    • The machine learning system has been trained on millions of annotations and can analyze entire images.
    • The results show that 1.5% of cancer cells are positive, which is significant but less than 50%.
    • The supporting results reveal that almost 12% of immune cells are positive.
    • Using algorithms can make pathologists more accurate, reproducible, and efficient.

    Improving Efficacy of Treatments with AI

    20:09 - 27:10

    • Measuring the efficacy of treatments through multiple biopsies over time can be improved with AI.
    • AI can analyze quantitative features before and after treatment in a scalable way.
    • This system can help assess the effectiveness of new drugs for diseases like NASH.
    • PDL1 protein expression in lung cancer patients determines treatment recommendations.
    • PDL1 inhibits the immune response to tumors, but drugs like Keytruda or Optivo can counteract this inhibition.
    • These drugs enable the immune system to work against cancer cells and improve survival rates.
    • Using algorithms can make pathologists more accurate, reproducible, and efficient.

    Advancements in Cancer Detection and Prevention

    26:52 - 33:49

    • Cancer treatments and early detection have made most cancers treatable now, unlike in the past where a cancer diagnosis was often a death sentence.
    • In the next 10 years, cancer is expected to become more manageable with advancements in early detection and risk factor prevention.
    • Cellular replication errors and environmental factors contribute to the development of cancer, making it more fundamental than rare events like plane crashes.
    • Preventing obesity and promoting exercise can have a significant impact on reducing cancer risk.
    • Screening for cancer has shown some benefits in finding curable cases, but overall survival improvements are still being studied.
    • Screening can lead to unnecessary procedures and anxiety if false positives occur.
    • While some individuals may prefer to go on a 'goose chase' for potential cancer findings, considering the impact on the population as a whole is important.

    Data-Driven Projects in Healthcare

    33:26 - 40:05

    • Daniel Act, the Spotify founder, has invested in a scanning company that aims to cost $400
    • LinkedIn is a great platform for finding and hiring candidates
    • Pooling data from various sources could lead to significant advancements in healthcare
    • Different groups, including industry and government-funded initiatives, are investing in data-driven projects
    • There is a diversity of approaches being taken to tackle healthcare challenges
    • Market-based capitalism and top-down systems both have their merits in driving innovation
    • Data organization in healthcare should be top-down and centralized to have a significant impact.
    • Pathology images have become highly generalizable across diverse tasks due to the scale of data collected over seven years.

    The Future of Pathology and Diagnostics

    39:49 - 46:10

    • The use of cloud and digital technology is transforming the field of pathology and diagnostics for drug development and clinical care.
    • Advances in digitization and connectivity are creating an inexhaustible digital resource for learning without destroying tissue.
    • Connecting brains across the world is a huge platform advance.
    • Training new models with lower cost, higher accuracy, reproducibility, and predictability will transform the field.
    • Compounding innovation is occurring in parallel with advancements in microscopes, electronics, smartphones, laptops, and video games.
    • Access to expert diagnoses is limited globally, but high-quality models deployed through the cloud can democratize access to expertise.
    • Digital technology and AI becoming more ubiquitous will unlock this process of democratizing access to expertise.
    • Working on meaningful projects like improving pathology diagnostics is more important than optimizing ads on platforms like Facebook.

    The Speaker's Unique Perspective

    52:36 - 53:33

    • The speaker has a unique position of being able to connect with both entrepreneurs and investors.
    • Through conversations, the speaker is able to draw connections between different industries and companies.
    • This ability is likened to the scene in The Hangover where Zach Galifianakis does math and symbols fly over his head.
    • The episode concludes with a mention of Pathai.com and a farewell message.
    1