Kakao Healthcare Corp. launched an artificial intelligence (AI)-powered mobile application called PASTA on Feb. 1 for continuous glucose monitoring (CGM)-based diabetes care. PASTA pairs with two types of CGM sensors, either from Seoul-based I-sens Inc.’s Caresens Air or San Diego-based Dexcom Inc.’s G7 system.

To date, developers of digital health technologies have had little luck with coverage and reimbursement, but the picture is slowly improving. The U.K.’s National Institute for Health and Care Excellence (NICE) has endorsed two artificial intelligence (AI) algorithms for diagnosis of stroke, although this is limited to an evidence generation framework that puts the sponsors on the hook for generating the data that will allow unlimited use in the U.K.’s trusts.

The U.S. FDA’s device center has at times struggled to make the volume of hires under the reigning Medical Device User Fee Agreement (MDUFA), but that wasn’t a problem in fiscal year 2023.

A team of inventors predominantly based in the state of New York are seeking patent protection for an earbud system that integrates real-time electroencephalography (EEG) monitoring with audio playback. Using specialized silicone electrodes, it captures and interprets EEG data, serving as a passive and continuous tool for screening neurological conditions.

Quibim SL recently launched its artificial intelligence (AI)-based software QP-Brain, which is designed to detect early-stage neurodegenerative diseases, after it received U.S. FDA 510(k) clearance, as well as CE and UKCA marks from the EU and U.K. regulators, respectively. QP-Brain quantifies and presents data from patients’ brain images to help clinicians with early diagnosis and treatment strategies for diseases such as Alzheimer’s, multiple sclerosis and dementia.

Syntheticgestalt Ltd. and Enamine Ltd. have begun a joint effort to create a suite of artificial intelligence (AI) models that will enable the generation of synthetically accessible biologically active compounds with optimized physicochemical and ADME/Tox properties.

Researchers have reported that the predictive abilities of a machine learning algorithm trained using best practices on a large clinical dataset did not generalize beyond the data that was used to train it. The algorithm was able to predict, to a degree, which individual patients would benefit from the medication when the patients were from the dataset the algorithm was trained on. But when it was supposed to predict who would benefit in clinical cohorts that were not part of the training, it performed no better than chance.

The first patenting from Mhealthcare Inc. describes a patient examination table or bed equipped with a variety of sensors, data from which may be analyzed with trained machine learning models to facilitate risk assessment and diagnosis of non-neurotypical developmental conditions such as autism in infants and young children by predicting cognitive, behavioral, social and developmental outcomes as early as the first three months of life. It is also claimed that the table may be used to diagnose epilepsy and Alzheimer’s disease.

If we unraveled the DNA of the 46 chromosomes of a single human cell, it would barely measure 2 meters. If we did the same with the rest of the body, if we aligned the 3 billion base pairs of its 5 trillion cells, we could travel the distance from the Earth to the Sun more than 100 times. It seems unreachable. However, that is the unit of knowledge of the large sequencing projects achieved in 2023.