Month: April 2023

FDA offers guidance on long-term neonate product studies: The FDA released draft guidance outlining recommendations for sponsors when conducting long-term neurodevelopmental studies to support the safety of medical products such as drugs, biologics or devices intended for use in neonates. According to the guidance, “although a single-arm study may be useful for collecting some types of safety information, the absence of a concurrent control arm (placebo or active comparator) will generally make clear interpretation of the results difficult, if not impossible.”

Large language model identifies people with Alzheimer’s: GPT-3, the large language model behind ChatGPT, achieved 80% accuracy in identifying people with Alzheimer’s Disease (AD) based on software-transcribed 10-second audio clips of them describing pictures in a standard dementia test. People diagnosed with AD repeated themselves, went off-topic, did not finish their thoughts and referred to objects vaguely, researchers reported in PLOS Digital Health.

3M launches new adhesive tape with 28-day wear time: 3M has released Medical Tape 4578, its new medical adhesive that has liner-free stability and can be worn on the skin for up to 28 days. The tool can also be stored for up to a year.

Artificial Intelligence (AI) predicts autism risk in bases based on Electronic Health Record (HER) data: Researchers used EHR data from 45,080 babies up to a year old to train an autism prediction model that achieved high sensitivity, specificity and positive predictive value, according to a study published in JAMA Network Open. The algorithm “could be integrated with caregiver surveys to improve the accuracy of early autism screening,” researchers wrote.

Smartphone app might detect onset of stroke: A smartphone machine learning algorithm called FAST AI detects facial asymmetry, arm weakness and speech changes associated with the onset of a stroke. Studies of FAST AI conducted at four stroke centers in Bulgaria found the app was accurate, but some experts say false negatives could be dangerous, and the technology should be used to prompt users to seek medical attention quickly, not to rule out a stroke.

Machine-learning method offers view of placental compartments: A machine-learning method for automated MRI scan processing has been developed by NIH-funded researchers to provide a detailed view of compartments in the placenta, including intervillous space and placental tissue and vessels. The technique is being studied for the potential to aid in early monitoring of placental health to identification and prevention of complications in pregnancy.

Immunexpress tool can flag COVID-19 patients needing hospitalization: A study that assessed SeptiCyte RAPID from Immunexpress has validated the test as a triaging tool to identify disease severity and predict hospitalization or Intensive Care Unit (ICU) care needs of patients with COVID-19. The tool was able to identify critical and severe patients when used within 48 hours of intake.

FDA awarded fewer breakthrough device tags in 2022: The annual report from the FDA’s Center for Devices and Radiological Health (CDRH) shows a nearly 37% drop in devices awarded breakthrough designations. Medical devices awarded breakthrough tags dropped to 135 from 213 in 2021, while devices with breakthrough status that were given marketing authorization increased from 13 to 19.