AI in Medicine - curated summaries making complex issues easy to understand

Mike Rawson
  • 0.0 (0)
  • Medicine
  • Weekly

AI in Medicine - Smart Summaries is a daily podcast for clinicians, health-system leaders and medtech, pharma and digital-health teams. Each episode unpacks one important public document on artificial intelligence in healthcare, such as FDA, EMA, MHRA, WHO or NHS guidance, OECD and World Economic Forum reports, or an open-access peer-reviewed review, in a clear 15 to 20 minute conversation. Topics include clinical AI and decision support, medical imaging and radiology AI, large language models (LLMs) and generative AI in clinical practice, AI-enabled medical devices and their regulation, the EU AI Act, drug discovery and clinical trials, digital health and wearables, health-system adoption, patient safety, bias and liability. The source of every episode is linked in the show notes, and full transcripts are at ai-in-medicine-podcast.vercel.app. The hosts' voices are AI-generated. For information only, not medical advice.

  1. 2h ago

    Company Spotlight: GuideAI VAOT - Triage AI for Peripheral Arteries

    We unpack the FDA 510(k) summary for GuideAI's Vascular Assist Occlusion Triage (VAOT) software. We explore its clinical targets in the lower extremities, the technical workflow, and how it achieved clearance using a stroke-triage AI as its predicate device. Key points GuideAI's VAOT is a Class II triage tool for detecting total arterial occlusion in the lower extremities, achieving a sensitivity of 0.915.The software operates entirely in parallel with standard-of-care, sending notifications via DICOM Secondary Captures.VAOT secured FDA clearance by demonstrating substantial equivalence to Aidoc's Briefcase LVO Triage, a device used for stroke detection in the brain.Performance testing was retrospective, using a 317-case dataset enriched with 21 positive aortic occlusions. Source: FDA 510(k) summary K260729: Vascular Assist Occlusion Triage (VAOT) (v1.0) (GuideAI Health) - U.S. Food and Drug Administration, 2026 This episode is an AI-generated conversation summarising a public document; the hosts' voices are synthetic. It is for information only and is not medical advice. Always refer to the original source. Company: GuideAI Health (guideaihealth.com) · Product: Vascular Assist Occlusion Triage (VAOT) (v1.0) · FDA 510(k) K260729 Full transcript: https://ai-in-medicine-podcast.vercel.app/episodes/company-spotlight-guideai-vaot-triage-ai-for-peripheral-arteries-a66eea

  2. 19h ago

    Company Spotlight: Caristo CaRi-Plaque - 0.999 Correlation with Remote Operators

    In this episode, we unpack the FDA 510(k) summary for Caristo Diagnostics' CaRi-Plaque software. We explore how this web-based device leverages remote human operators and threshold-based segmentation to achieve remarkably high correlations with expert cardiologists, and discuss the unusual demographic estimations used in its clinical validation. Key points CaRi-Plaque is a web-based, Class II diagnostic reading software intended for analyzing cardiac CT data to determine coronary plaques and luminal stenoses.Unlike its machine-learning predicate, CaRi-Plaque uses threshold-based segmentation and relies on remote trained operators in regional data centers rather than local clinical workstations.Clinical validation on 117 subjects showed incredibly high Pearson Correlation Coefficients against expert ground truth, including 0.999 for Calcified Plaque Volume.The validation dataset was heavily skewed toward men (85 men to 32 women) and utilized estimated local census data to determine patient ethnicity for over half the subjects.The FDA label warns that interpolated data generated by the software may mask very small pathologies, emphasizing that the tool cannot replace a qualified practitioner's judgment. Source: FDA 510(k) summary K242240: CaRi-Plaque (Caristo Diagnostics) - U.S. Food and Drug Administration, 2025 This episode is an AI-generated conversation summarising a public document; the hosts' voices are synthetic. It is for information only and is not medical advice. Always refer to the original source. Company: Caristo Diagnostics · Product: CaRi-Plaque · FDA 510(k) K242240 Full transcript: https://ai-in-medicine-podcast.vercel.app/episodes/company-spotlight-caristo-cari-plaque-0-999-correlation-with-remote-op-cbe30a

  3. 20h ago

    When Heart Fat Becomes a Thermometer: AI, CT Scans, and the Future of Plaque Detection

    Routine coronary CT scans are getting an AI-driven upgrade. We explore a paper detailing how artificial intelligence analyzes pericoronary adipose tissue to act as a dynamic biosensor, detecting the invisible vascular inflammation that causes heart attacks before arteries are fully blocked. Key points Coronary computed tomography angiography is a standard test, but fails to identify non-obstructive plaques that often rupture and cause acute cardiac events.Pericoronary adipose tissue acts as a non-invasive "thermometer" of vascular inflammation, changing its structural and lipid composition when adjacent vessels are inflamed.The Fat-Attenuation Index Score uses AI to standardize these changes across different CT scanners, reading shifts in tissue density that indicate higher inflammatory risk.Deep learning models, specifically convolutional neural networks and 3D U-Net architectures, are automating the complex segmentation of this fat, achieving near-expert accuracy.Therapeutic interventions, including statins and biologics, have been shown to measurably reduce Fat-Attenuation Index scores, proving this metric can track treatment response dynamically.Despite immense promise, AI explainability remains a hurdle, with many models functioning as "black boxes" that complicate regulatory approval and clinical trust. Source: Pericoronary adipose tissue imaging: quantitative assessment, artificial intelligence integration, and therapeutic modulation - The British journal of radiology, 2026 (CC BY) This spot is available. Reach clinicians, health-system leaders and medtech and pharma teams following AI in medicine. Sponsor the show This episode is an AI-generated conversation summarising a public document; the hosts' voices are synthetic. It is for information only and is not medical advice. Always refer to the original source. Authors: Leanne Eveson, Ikboljon Sobirov, Kenneth Chan · DOI: 10.1093/bjr/tqag040 Full transcript: https://ai-in-medicine-podcast.vercel.app/episodes/when-heart-fat-becomes-a-thermometer-ai-ct-scans-and-the-future-of-pla-cd7f26

  4. 1d ago

    Unpacking Health Canada's Roadmap for Machine Learning Medical Devices

    Health Canada has released its pre-market guidance for machine learning-enabled medical devices, outlining how developers must validate, document, and license AI tools. Maya and Sam break down the evidence thresholds, clinical validation standards, and the game-changing Predetermined Change Control Plan that allows pre-authorized software updates. Key points The guidance applies to Class II, III, and IV machine learning-enabled medical devices, spanning both standalone software and integrated systems.Health Canada introduces the Predetermined Change Control Plan, enabling manufacturers to pre-authorize planned changes to address model degradation without triggering a significant change amendment.Developers must integrate Sex and gender-based analysis plus across the device lifecycle, ensuring data reflects the diverse Canadian population.Class III and IV devices require clinical validation studies using populations entirely independent from the development and tuning data.Risk management must address algorithmic failures like overfitting and underfitting, as well as human-interaction risks like automation bias and alarm fatigue.Transparency is mandated through labelling, with the guidance suggesting a structured summary like a model card or data card for end-users. Source: Pre-market guidance for machine learning-enabled medical devices - Health Canada, 2026 This spot is available. Reach clinicians, health-system leaders and medtech and pharma teams following AI in medicine. Sponsor the show This episode is an AI-generated conversation summarising a public document; the hosts' voices are synthetic. It is for information only and is not medical advice. Always refer to the original source. Full transcript: https://ai-in-medicine-podcast.vercel.app/episodes/unpacking-health-canada-s-roadmap-for-machine-learning-medical-devices-ab348e

  5. 2d ago

    Company Spotlight: Imvaria Fibresolve - AI for Lung Fibrosis with a Built-In Update Plan

    We unpack the FDA 510(k) summary for Imvaria's Fibresolve with PCCP. This software-only device analyzes lung CT scans as an adjunct for diagnosing idiopathic pulmonary fibrosis prior to invasive testing. We explore how its Predetermined Change Control Plan allows the manufacturer to update and re-train the AI model over time without requiring new regulatory clearance. Key points Fibresolve is a Class II software-only device that uses an ensemble 3D deep learning model to analyze DICOM-compliant lung CT scans for suspected interstitial lung disease.It acts as an adjunct in the diagnosis of idiopathic pulmonary fibrosis (IPF) prior to invasive testing, specifically for patients strictly > 22 years old.The clearance is based on substantial equivalence to the original Fibresolve (DEN220040), but the summary notably does not provide specific clinical dataset sizes or performance metrics.The clearance includes a Predetermined Change Control Plan (PCCP), allowing Imvaria to modify model architecture, introduce new training data, and incorporate ancillary inputs like age, sex, and pulmonary function tests without submitting a new 510(k). Source: FDA 510(k) summary K252041: Fibresolve (with PCCP) (Imvaria) - U.S. Food and Drug Administration, 2025 This episode is an AI-generated conversation summarising a public document; the hosts' voices are synthetic. It is for information only and is not medical advice. Always refer to the original source. Company: Imvaria (imvaria.com) · Product: Fibresolve (with PCCP) · FDA 510(k) K252041 Full transcript: https://ai-in-medicine-podcast.vercel.app/episodes/company-spotlight-imvaria-fibresolve-ai-for-lung-fibrosis-with-a-built-ea3e45

  6. 2d ago

    Company Spotlight: Cercare Medical Virtual Expert - Mapping Brain Tumors and Blood Flow

    In this episode, we unpack the FDA 510(k) summary for Cercare Medical's CMN Capillary Function with Oncology Virtual Expert. We explore how this software combines structural tumor volumes with capillary blood flow data, the reality of its 0.706 and 0.822 Dice similarity scores, and the strict scanning rules required for longitudinal tracking. Key points The FDA cleared Cercare Medical's Oncology Virtual Expert as a Class II decision-support tool for adult patients with pathologically proven brain gliomas, WHO Grade II through IV.Unlike its cloud-based predicate device, the software runs locally on standard off-the-shelf computers or virtual platforms like VMware.Testing on a 100-patient dataset showed a mean Dice similarity coefficient of 0.706 for T1 enhancing regions and 0.822 for T2/FLAIR hyperintensity regions.The system automatically extracts perfusion data, such as tissue blood volume and flow, directly from the delineated tumor sub-components.Device labeling strictly instructs that longitudinal volumetric comparisons must be performed using the same scanner model and consistent acquisition parameters to minimize variability. Source: FDA 510(k) summary K260316: CMN Capillary Function with Oncology Virtual Expert (Cercare Medical) - U.S. Food and Drug Administration, 2026 This episode is an AI-generated conversation summarising a public document; the hosts' voices are synthetic. It is for information only and is not medical advice. Always refer to the original source. Company: Cercare Medical (cercare-medical.com) · Product: CMN Capillary Function with Oncology Virtual Expert · FDA 510(k) K260316 Full transcript: https://ai-in-medicine-podcast.vercel.app/episodes/company-spotlight-cercare-medical-virtual-expert-mapping-brain-tumors-a3beb9

  7. 3d ago

    Beyond Time Saved: The Hidden Risks of Evaluating AI Scribes on Efficiency Alone

    Health systems are rushing to deploy ambient AI scribes to combat clinician burnout. This episode unpacks a new framework explaining why evaluating these tools solely on minutes saved is a dangerous trap, and how to measure communication, narrative quality, and equity instead. Key points Physician burnout represents a significant workforce crisis, driving rapid adoption of AI scribes.A meta-analysis of 23 studies showed only a moderate pooled reduction in documentation workload.Across 6 major real-world implementations, none systematically assessed patient-clinician communication, narrative fidelity, or equity.AI documentation errors can create silent restructuring where a missed red-flag symptom quietly redirects downstream diagnostic reasoning.Evaluating AI scribes requires valid instruments like CAHPS and PDQI-9, alongside AI-off probes for trainees and equity analyses for multilingual encounters. Source: Beyond Time Saved: A Framework for Comprehensive Evaluation of AI Clinical Documentation - JMIR medical informatics, 2026 (CC BY) This spot is available. Reach clinicians, health-system leaders and medtech and pharma teams following AI in medicine. Sponsor the show This episode is an AI-generated conversation summarising a public document; the hosts' voices are synthetic. It is for information only and is not medical advice. Always refer to the original source. Authors: Ekstrom Victoria Sze Min · DOI: 10.2196/86091 Full transcript: https://ai-in-medicine-podcast.vercel.app/episodes/beyond-time-saved-the-hidden-risks-of-evaluating-ai-scribes-on-efficie-0f158d

  8. 3d ago

    Company Spotlight: HistoSonics Planning Tool - AI Mapping for Histotripsy

    We unpack the FDA 510(k) summary for the HistoSonics Planning Tool, cleared in May 2026. This episode explores how the standalone software uses machine learning to segment organs and map acoustic pathways for histotripsy procedures. We dive into the training data, the validation subsets, and what this clearance means for clinical workflows. Key points The FDA cleared the HistoSonics Planning Tool as a Class II medical image management and processing system based on substantial equivalence to a 2019 microwave ablation visualization device.The software uses static machine learning models to automatically segment the liver, kidneys, lungs, and ribs in CT and MRI scans to help plan acoustic pathways.Performance testing showed the liver segmentation model achieved a mean Dice score of 93%, exceeding the target of 70% or higher in 70% of cases.The validation relied entirely on retrospective datasets, including subsets of 10 patients from larger public databases, with some imaging data dating back to the period from 1991 to 1995.Because it is a planning tool, the device was cleared through software testing and did not require new animal or human clinical studies. Source: FDA 510(k) summary K252947: HistoSonics® Planning Tool (HistoSonics) - U.S. Food and Drug Administration, 2026 This episode is an AI-generated conversation summarising a public document; the hosts' voices are synthetic. It is for information only and is not medical advice. Always refer to the original source. Company: HistoSonics (histosonics.com) · Product: HistoSonics® Planning Tool · FDA 510(k) K252947 Full transcript: https://ai-in-medicine-podcast.vercel.app/episodes/company-spotlight-histosonics-planning-tool-ai-mapping-for-histotripsy-752a1c

See All (152)

About

AI in Medicine - Smart Summaries is a daily podcast for clinicians, health-system leaders and medtech, pharma and digital-health teams. Each episode unpacks one important public document on artificial intelligence in healthcare, such as FDA, EMA, MHRA, WHO or NHS guidance, OECD and World Economic Forum reports, or an open-access peer-reviewed review, in a clear 15 to 20 minute conversation. Topics include clinical AI and decision support, medical imaging and radiology AI, large language models (LLMs) and generative AI in clinical practice, AI-enabled medical devices and their regulation, the EU AI Act, drug discovery and clinical trials, digital health and wearables, health-system adoption, patient safety, bias and liability. The source of every episode is linked in the show notes, and full transcripts are at ai-in-medicine-podcast.vercel.app. The hosts' voices are AI-generated. For information only, not medical advice.

Information