Description
Researchers Discover Bias in AI Models That Analyze Pathology Samples | Harvard Medical School
Hosted on Ausha. See ausha.co/privacy-policy for more information.
Transcription
- Nancy Johnson
Something as simple as a blood test can open up a hospital bed. Hear from a CAP expert who is easing capacity issues at his hospital. That story and more coming up next. This is Path News Network Daily Edition from the College of American Pathologists. I'm Nancy Johnson. It's Monday, January 26th. Hospitals across the country face crowded emergency rooms and full inpatient floors. It's the lab that's keeping patients moving through the system. At Newton Wellesley Hospital in Massachusetts, turnaround time for routine lab tests has become a key focus, especially when it comes to discharges. The CAP's Dr. Michael Miselec chairs pathology there and says the pressure on turnaround time is constant.
- Michael Miselec
High capacity and high census, like we are here now and many institutions across the country are really at a high level of inpatients both in the ED and looking for beds on the floors and also looking for discharges.
- Nancy Johnson
Dr. Misalek says the biggest impact often comes from everyday testing.
- Michael Miselec
It's not just esoteric blood tests but it's routine morning labs. routine chemistries, CBCs, things like that, that we run every day high volume. And it's these little things that make a big difference to patient care and the hospital bottom line.
- Nancy Johnson
Hospital leaders closely track length of stay, sometimes down to fractions of a day, because it directly affects costs and reimbursements. Dr. Misalek says the lab was asked to make sure test results never slowed down a discharge.
- Michael Miselec
We've worked with both medical and nursing leadership to get lists every day of which patients are slated for possible discharge on each unit so that our phlebotomy teams can go to those high priority areas, draw the patient's blood, get it down to the lab so that we can run it and get results turned around quickly.
- Nancy Johnson
Morning labs are especially critical. in guiding care decisions and discharge planning. Newton Wellesley now holds twice-daily capacity meetings with lab reporting, helping free up beds and improve patient flow. The Association for Diagnostics and Laboratory Medicine has launched a first-of-its-kind Data Science and Laboratory Medicine Certificate Program for clinical lab professionals and pathologists. The program is designed to help participants use real-world healthcare data to improve testing algorithms, reduce errors, and boost lab efficiency. ADLM leaders say clinical labs generate billions of test results every year, but until now there has been no comprehensive training focused on applying data science in lab medicine. The new certificate is designed to equip professionals with in-demand analytic skills in a field increasingly driven by data. The digital pathology market is expected to grow from $2 billion in 2026 to $3.2 billion by 2031, driven by artificial intelligence and remote diagnostic technologies. A mortar intelligence report says AI-powered tools are helping pathologists detect and classify diseases more accurately, reduce workload, and speed up oncology diagnoses. North America, Europe, and the Asia Pacific region are leading this growth with hospitals and labs adopting whole slide imaging and digital workflows. is transforming diagnostics, improving collaboration among pathologists, and increasing access to expert care worldwide. And finally, a new study finds that artificial intelligence models used to analyze pathology slides for cancer diagnosis can perform unevenly across different demographic groups. Researchers at Harvard Medical School identified several causes of the bias and developed a tool called FAIRPATH that significantly reduces disparities in AI performance. The senior author of the study says the findings highlight the importance of checking pathology AI for bias to ensure accurate and equitable care. The team hopes that FAIRPATH and similar approaches can help create AI models that improve cancer diagnosis for patients of all ages, races, and genders. That's all for today. Be sure to check the show notes for more information on today's stories. Watch your email inbox for more news like this in the CAP's advocacy newsletter every Tuesday and our weekly edition newsletter on Thursdays. We're back tomorrow at 5 a.m. Eastern for another episode of The Daily Edition. I'm Nancy Johnson. Have a great day.
Description
Researchers Discover Bias in AI Models That Analyze Pathology Samples | Harvard Medical School
Hosted on Ausha. See ausha.co/privacy-policy for more information.
Transcription
- Nancy Johnson
Something as simple as a blood test can open up a hospital bed. Hear from a CAP expert who is easing capacity issues at his hospital. That story and more coming up next. This is Path News Network Daily Edition from the College of American Pathologists. I'm Nancy Johnson. It's Monday, January 26th. Hospitals across the country face crowded emergency rooms and full inpatient floors. It's the lab that's keeping patients moving through the system. At Newton Wellesley Hospital in Massachusetts, turnaround time for routine lab tests has become a key focus, especially when it comes to discharges. The CAP's Dr. Michael Miselec chairs pathology there and says the pressure on turnaround time is constant.
- Michael Miselec
High capacity and high census, like we are here now and many institutions across the country are really at a high level of inpatients both in the ED and looking for beds on the floors and also looking for discharges.
- Nancy Johnson
Dr. Misalek says the biggest impact often comes from everyday testing.
- Michael Miselec
It's not just esoteric blood tests but it's routine morning labs. routine chemistries, CBCs, things like that, that we run every day high volume. And it's these little things that make a big difference to patient care and the hospital bottom line.
- Nancy Johnson
Hospital leaders closely track length of stay, sometimes down to fractions of a day, because it directly affects costs and reimbursements. Dr. Misalek says the lab was asked to make sure test results never slowed down a discharge.
- Michael Miselec
We've worked with both medical and nursing leadership to get lists every day of which patients are slated for possible discharge on each unit so that our phlebotomy teams can go to those high priority areas, draw the patient's blood, get it down to the lab so that we can run it and get results turned around quickly.
- Nancy Johnson
Morning labs are especially critical. in guiding care decisions and discharge planning. Newton Wellesley now holds twice-daily capacity meetings with lab reporting, helping free up beds and improve patient flow. The Association for Diagnostics and Laboratory Medicine has launched a first-of-its-kind Data Science and Laboratory Medicine Certificate Program for clinical lab professionals and pathologists. The program is designed to help participants use real-world healthcare data to improve testing algorithms, reduce errors, and boost lab efficiency. ADLM leaders say clinical labs generate billions of test results every year, but until now there has been no comprehensive training focused on applying data science in lab medicine. The new certificate is designed to equip professionals with in-demand analytic skills in a field increasingly driven by data. The digital pathology market is expected to grow from $2 billion in 2026 to $3.2 billion by 2031, driven by artificial intelligence and remote diagnostic technologies. A mortar intelligence report says AI-powered tools are helping pathologists detect and classify diseases more accurately, reduce workload, and speed up oncology diagnoses. North America, Europe, and the Asia Pacific region are leading this growth with hospitals and labs adopting whole slide imaging and digital workflows. is transforming diagnostics, improving collaboration among pathologists, and increasing access to expert care worldwide. And finally, a new study finds that artificial intelligence models used to analyze pathology slides for cancer diagnosis can perform unevenly across different demographic groups. Researchers at Harvard Medical School identified several causes of the bias and developed a tool called FAIRPATH that significantly reduces disparities in AI performance. The senior author of the study says the findings highlight the importance of checking pathology AI for bias to ensure accurate and equitable care. The team hopes that FAIRPATH and similar approaches can help create AI models that improve cancer diagnosis for patients of all ages, races, and genders. That's all for today. Be sure to check the show notes for more information on today's stories. Watch your email inbox for more news like this in the CAP's advocacy newsletter every Tuesday and our weekly edition newsletter on Thursdays. We're back tomorrow at 5 a.m. Eastern for another episode of The Daily Edition. I'm Nancy Johnson. Have a great day.
Share
Embed
You may also like