|
How Are Doctors Using AI? | Is AI a Health Hazard?
December 9, 2024
|
|
|
|
|
Together with
|
|
|
|
“Back-end AI is being deployed widely, Front-end AI not so much.”
|
|
Herman Oosterwijk, in an analysis of AI developments at last week’s RSNA 2024.
|
|
|
Medical imaging technology like PET and MRI are helping physicians diagnose and manage patients being treated with the new generation drugs for Alzheimer’s disease. In this Imaging Wire Show, we talked to Kevin Ulm and Saurabh Sharma of Siemens Healthineers about these developments.
|
|
|
|
How are healthcare providers who have adopted AI really using it? A new Medscape/HIMSS survey found that most providers are using AI for administrative tasks, while medical image analysis is also one of the top AI use cases.
AI has the potential to revolutionize healthcare, but many industry observers have been frustrated with the slow pace of clinical adoption.
- Implementation challenges, regulatory issues, and lack of reimbursement are among the reasons keeping more healthcare providers from embracing the technology.
But the Medscape/HIMSS survey shows some early successes for AI … as well as lingering questions.
- Researchers surveyed a total of 846 people in the U.S. who were either executive or clinical leaders, practicing physicians or nurses, or IT professionals, and whose practices were already using AI in some way.
The top four tasks for which AI is being used were administrative rather than clinical, with image analysis occupying the fifth spot …
- Transcribing patient notes (36%).
- Transcribing business meetings (32%).
- Creating routine patient communications (29%).
- Performing patient record-keeping (27%).
- Analyzing medical images (26%).
The survey also analyzed attitudes toward AI, finding …
- 57% said AI helped them be more efficient and productive.
- But lower marks were given for reducing staff hours (10%) and lowering costs (31%).
- AI got the highest marks for helping with transcription of business meetings (77%) and patient notes (73%), reviewing medical literature (72%), and medical image analysis (70%).
The findings track well with developments at last week’s RSNA 2024, where AI algorithms dedicated to non-clinical tasks like radiology report generation, scheduling, and operation analysis showed growing prominence.
- Indeed, many AI developers have specifically targeted the non-clinical space, both because commercialization is easier (FDA authorization is not typically needed) and because doctors often say they need more help with administrative rather than clinical tasks.
The Takeaway
While it’s easy to be impatient with AI’s slow uptake, the Medscape/HIMSS survey shows that AI adoption is indeed occurring at medical practices. And while image analysis was radiology’s first AI use case, speeding up workflow and administrative tasks may end up being the technology’s most impactful application.
|
|
|
Discover AI Apps Curated by Body Regions
Discover how Calantic Digital Solutions by Bayer supports you in tackling radiology challenges through multiple stages of the patient’s journey by offering AI solutions curated for specific body regions. Learn more about Calantic today.
|
|
AI for Detecting Fractures on X-Rays
Missed fractures occur all too frequently and are a significant cause of patient morbidity. Learn how U.K. researchers are using Gleamer’s BoneView algorithm to discover how AI can help radiologists improve their diagnostic accuracy for fracture detection on X-rays.
|
|
Feel the Freedom of Helium-Free MRI
Lift limitations and experience MRI excellence with Philips BlueSeal, the industry’s lightest, vent pipe-free, high-performance, helium-free 1.5T scanner. Save on helium and energy costs, achieve precise AI-enhanced diagnoses, enjoy faster scans, and optimized workflows. Learn more today..
|
|
- ‘Risks of AI’ Tops ECRI Hazard List: Not everyone is excited about AI’s potential impact on healthcare. Healthcare research firm ECRI named “risks with AI-enabled technologies” as the number one threat to healthcare providers in its annual top 10 health technology hazards list. ECRI cited misleading AI results like hallucinations as one major threat, as well as bias when AI is used with different patient populations. On the positive side, no other imaging technologies landed on the list, unlike in past years.
- Herman O’s RSNA Review: If you’re looking for a great overview of AI at RSNA 2024, check out Herman Oosterwijk’s review of last week’s meeting. Oosterwijk estimates that AI companies occupied one-third of McCormick Place’s South Hall, and while interest in AI was red hot, he notes many of the technology’s shortcomings, including a lack of transparency, shaky clinical evidence, and cumbersome implementation. But AI still has enormous potential if it can give time back to clinicians to spend on patient care such as by automating tedious tasks.
- LLMs Produce Patient Reports: Sure, large language models can be used to create radiology reports for patients, but how good are they? Researchers put LLMs to the test in a Thursday poster presentation at RSNA 2024, finding that for 401 abdominal CT exams, the WizardLM algorithm created patient reports that gave a clear impression summary for 81% of cases, with higher readability (8th grade level). But the reports missed key information in 3.5% and 10% included AI hallucinations, highlighting the importance of human oversight to avoid inaccuracies.
- AI Boosts Breast Screening Outreach: In another novel use of AI from the same RSNA session, researchers showed how an AI algorithm boosted breast screening compliance by directing patient outreach based on AI-generated scores. The scores were used to adjust outreach cadence over three months with text, voice calls, or voicemail at 272 imaging centers starting in 2019. In 6.6M patients, screening compliance was up 8% over four years after the program started, with the biggest jump among Black women (+12%), historically one of the most underserved groups.
- Better Breast Cancer Detection with AI: Another RSNA 2024 presentation from Sunday found that Kheiron Medical’s Mia algorithm helped improve cancer detection rates while reducing workload in a Scottish breast screening program. In a population of 10.9k women, researchers tested 17 different AI workflows, with AI used as a safety net to standard double reading by flagging suspicious cases. Additional arbitration took place for 12% of cases, but the cancer detection rate rose by 1 cancer per 1k cases, with workload savings of 31%. Kheiron was acquired by DeepHealth in October.
- DeepHealth’s RSNA Developments: At RSNA 2024, DeepHealth took a step toward expanding its recently launched SmartTechnology AI portfolio by announcing an agreement to provide its SmartSonography solution to Siemens Healthineers. SmartSonography includes workflow optimization features like remote scanning as well as a cloud-native enterprise viewer. Separately, DeepHealth partnered with platform developer CARPL.ai to create an AI control system to monitor AI performance, and the company received FDA clearance for use of its SmartMammo AI with GE HealthCare’s Pristina mammography system. Finally, DeepHealth launched Diagnostic Suite and TechLive at RSNA 2024.
- AI Boosts Underperforming Rads: An AI algorithm that autonomously ruled out DBT exams at low suspicion for breast cancer helped boost underperforming radiologists in a Tuesday morning RSNA poster session. In a study of 26 radiologists reading 155k screening exams, researchers from DeepHealth found that AI increased the cancer detection rate (15%), recall rate (8%), and PPV1 (7%), and boosted CDR to the target of 2.5 per 1k exams. Setting the autonomous rule-out threshold to 60% brought 92% of radiologists into an acceptable range for CDR and recalls.
- Prenuvo Touts RSNA Studies: Whole-body screening provider Prenuvo highlighted results of an RSNA 2024 study that linked type 2 diabetes to premature brain aging. Researchers performed whole-body MRI scans on 650 people and analyzed 96 brain regions, finding that people with diabetes had lower brain volume, especially in the occipital lobe. Other Prenuvo studies included research on normative musculoskeletal aging curves for lower limbs and a study on the impact of body composition on brain health that linked lower skeletal muscle mass percentage to lower brain volume.
- Gleamer Gets Funding, Lands Konica Minolta: French AI developer Gleamer secured €4.5M ($4.8M) from the French government to accelerate its OncoView project to develop oncology AI applications. Gleamer is expanding into oncology from its core focus on musculoskeletal AI, and is enhancing its Gleamer Copilot solution to support interpretation of oncology CT. Separately, Gleamer is providing its BoneView AI fracture solution to Konica Minolta Healthcare Americas, which will offer it alongside its DR systems.
- FDA Issues AI Post-Market Rules: The FDA last week issued its final guidance document on predetermined change control plans for AI-enabled medical devices. The guidance creates a regulatory framework for AI developers to monitor their products after they’ve been approved and make adjustments to accommodate the fact that AI performance changes as algorithms are exposed to new data. The guidance applies to AI devices regulated through the 510(k), de novo, and PMA pathways.
- Cancer Screening Saves Lives: A new modeling study in JAMA Oncology documents the live-saving value of cancer screening for five cancers: breast, cervical, colorectal, lung, and prostate from 1975 to 2020. For breast cancer, researchers estimated that prevention, screening, and treatment advances saved 1M lives, with 25% of the benefit due to screening. For lung cancer, smoking cessation accounts for 98% of the mortality reduction (3.5M deaths averted) due to the low uptake of lung screening. The findings track with other recent studies on mammography’s contribution.
- Chest X-Ray AI Finds Nodules: Qure.ai’s qXR AI algorithm helped radiologists find lung nodules on chest X-rays in a study in Academic Radiology. Researchers used qXR to analyze chest radiographs of 300 patients from 40 U.S. hospitals, finding that AI improved mean AFROC of 15 readers (0.81 vs. 0.73) as well as case-level sensitivity (84% vs. 73%) while specificity was similar (72% vs. 71%). The study was funded by Qure and was used as part of its regulatory submission for qXR.
- Nanox Gets Broader FDA Indication: Digital X-ray developer Nanox received FDA 510(k) clearance to market its multi-source Nanox.ARC system for producing tomographic images for additional indications beyond what the company received in 2023. The new clearance covers indications like MSK, pulmonary, and intra-abdominal use, and Nanox believes it will boost the company’s commercialization efforts in the U.S., where Nanox.ARC is installed at sites in seven states.
- Lunit/Volpara RSNA Highlights: Lunit and Volpara continue to make progress in uniting operations brought together by acquisition in May 2024. At RSNA 2024, the companies said they are developing a unified ecosystem for early cancer detection, risk prediction, and autonomous AI, and demonstrated the combination of Lunit’s INSIGHT Risk and Volpara’s Risk Pathways solutions for breast cancer risk management. Lunit at RSNA also showed a prototype tool for autonomous chest X-ray report generation, based on foundation models.
- HeartFlow Plaque AI Matches IVUS: A retrospective subanalysis found that HeartFlow’s CCTA AI-based plaque quantification solution, AI-QCPA, performed similarly to intravascular ultrasound. Across 33 patients and 67 blood vessels, HeartFlow AI-QCPA agreed closely with IVUS measurements for vessel (ρ=0.94) and lumen volumes (ρ=0.97), with high agreement for total plaque volume (ρ=0.92), noncalcified plaque (ρ=0.91), and calcified plaque (ρ=0.87).
- Cleerly Adds $106M: Cardiac CT AI leader Cleerly concluded a $106M Series C extension round, following at least $279M in Series B and C funding raised since 2021. The funding was earmarked for expanding Cleerly’s commercial reach and continues the wave of cardiac CT AI venture capital alongside companies like Elucid ($80M in Series C) and HeartFlow ($838M in Series A – F).
- Center Chain Pays $5M in Kickback Case: An imaging center chain based in the southeastern U.S. paid federal and state authorities $5.3M to settle charges they gave kickbacks to physicians for patient referrals. American Health Imaging and its founder and former CEO Scott Arant paid the settlement to resolve charges that they gave referring physicians meals, tickets to sporting events, and other gifts for referrals from 2011 to 2019. The chain also allegedly paid physicians above fair market value to interpret scans they referred.
|
|
PACS that Scales to Fit Your Needs
IntelePACS from Intelerad streamlines your reading workflow from start to finish. With faster load times, centralized access, and integration with your EHR, it provides a smooth and efficient workflow your radiologists will prefer. Discover how IntelePACS can transform your radiology practice.
|
|
2 Questions about AI for Radiology Leaders
Are today’s radiology AI solutions solving the right problems? And are there other solutions available for AI of brain MRI? Read this article from SpinTech MRI to learn how its STAGE solution can optimize MRI utilization.
|
|
Transform Your Practice into a Multispecialty Powerhouse
Medality is the practice development platform that helps radiologists upskill in high-growth, advanced imaging areas. Request a demo today and discover how to transform your practice into a multispecialty powerhouse.
|
|
- Get Your Head Around AI for Neuroradiology: Check out the latest blog from Blackford on how advances in deep learning algorithms for neurology imaging are improving outcomes and easing the burden on radiologists.
- Digital Tools for Heart Failure: Clinicians have a growing array of digital tools for assessing patients with suspected heart failure. A new review article in Lancet Digital Health takes stock of some of the options, including echo AI tools like those from Us2.ai.
- AI and Cancer Screening: Cancer screening saves lives, but right now screening is limited to a few cancer types. That could change with AI, which opens new possibilities for earlier disease detection. Learn more in this article by DeepHealth clinical AI leader Greg Sorensen, MD.
- An End-to-End Solution for Viewing AI Output: Check out CARPL.ai’s FDA-cleared Universal AI View, an end-to-end solution for viewing, editing, and annotating AI outputs. It can be deployed on-premises or in the cloud to give you seamless interoperability. Book a meeting today to learn more.
- Stop Shipping Discs! By pivoting to a 100% digital fulfillment model for patient images and records, you can improve their experience while significantly reducing labor and shipping costs. Find out how on this page from Clearpath.
- The Clinical Value of Soft-Tissue Chest X-Ray: Soft-tissue techniques can improve the visibility and accuracy of chest X-ray. Learn about two important soft-tissue methods – bone suppression and dual-energy subtraction – in this white paper from Riverain Technologies.
- All Your Data in One Place: Mach7’s Vendor Neutral Archive (VNA) is a powerful data management and workflow orchestration technology built for the future. Learn how it can help you drive better patient care by centralizing all your data in one place.
- Keep Patients Engaged with Your Healthcare System: After using PocketHealth, 94% of patients were more confident about their healthcare experience. Learn how to increase follow-up adherence, improve communication, and provide screening tools to keep your patients on track with PocketHealth Patient Connect.
- The Road to Cloud-Based PACS: Radiology facilities are turning to cloud-based PACS like Visage’s Visage 7 to solve their medical image management needs. Learn about their experiences in this Imaging Wire Show with Amy Thompson of Signify Research and radiologist Marc Kohli, MD.
- An Enterprise Imaging Platform to Grow: WakeMed Health & Hospitals in North Carolina is home to award-winning chest pain centers and two nationally accredited, award-winning stroke centers. Find out how WakeMed turned to enterprise imaging solutions from AGFA HealthCare to transform the way their clinicians work.
- Turning Medical Imaging into Great Medicine: TeraRecon is continually innovating to bring the latest in advanced visualization capabilities to radiology. Intuition 4.7 includes a variety of new features, from structural heart workflow to the ability to trigger AI on demand. Learn more about them on this page.
- How McFarland Clinic Reduced Downtime: McFarland Clinic Health Ventures recently experienced a downtime occurrence when an HL7 interface went down. But thanks to a suite of medical imaging solutions from Merge by Merative, McFarland’s downtime lasted all of 15 minutes. Find out how they did it in this case study.
- Unprecedented Insights Made Possible with AI: With the largest normative dataset of whole-body imaging in the world, Prenuvo’s AI researchers partner with the best academic minds to understand – like never before – what “normal” aging means. Learn about their work today.
- Unlock Next-Generation AI with Foundation Models: Learn about Microsoft’s new family of cutting-edge multimodal medical imaging foundation models designed for healthcare organizations to test, fine-tune, and build tailored AI solutions specific to their needs, while minimizing extensive compute and data requirements.
- The Leader in Molecular Imaging: United Imaging’s uMI portfolio of solutions is designed to help you lead the way in molecular imaging. From digital PET/CT systems designed to stand the test of time to the cutting-edge uEXPLORER total-body PET scanner, discover the uMI difference today.
- Revolutionizing Medical Imaging Data Management: Enlitic has acquired Laitek, and the combination creates new possibilities to revolutionize medical imaging data management. Learn more about Laitek and how its advanced migration services can benefit your radiology practice.
|
|
|
|
|
