Developing Trust in Healthcare AI

Published by Daniel at November 1, 2022

Dr. Amy Sheng, a technical manager at Sino Biological, and Dr. Lurong Pan, founder, and CEO of the drug discovery platform Ainnocence, are working on a novel shared endeavor with high production and screening to improve the development process for potential universal vaccines. These universal vaccines provide extensive broad effectiveness against various strains of pathogens. From the specific case of SARS, scientists have experienced and understood how rapidly a virus mutates and evades immunity. Therefore, developing a vaccine that will continue to protect against any renewed or new versions of the virus is imperative. Universal vaccine development mainly focuses on the whole invariable portion of the virus or conserved region. Nonetheless, this region might be obscured by the virus’ ever-mutating domain, a phenomenon called immunodominance. Now, the Sino-labs are trying to make chimeric protein vaccines to make the conservative area of the virus more immunodominant, thus mounting a broader and more durable immune response. There are patterns in the evolutionary pathway of the viral species being discovered and able to predict the future mutation trends of a species. If AI can completely digitize the virus, its algorithm could analyze its evolutionary pattern to identify a standard immunogenetic sequence to contribute to the vaccine design process. The introduction of AI and computer science is bringing the development of such vaccines closer; with AI, researchers can conceive potential antibodies that target different viral strains and use the same algorithm to identify shared features among all mutant variants. This information would allow the creation of a targeted antibody to work even for future strains.(1)

Virtual Incision’s Miniaturized Robotic-Assisted Surgery Device Will Launch into Space in 2024

Virtual incision company, a medical technology manufacturer in Lincoln, Nebraska, had released its new pioneering medical device, the first miniaturized robotic-assisted surgery platform, called MIRA™. This device will test its space skills in a 2024 technology demonstration on a mission aboard the International Space Station (ISS). The opportunity is driven by a recent National Aeronautics and Space Administration (NASA) grant. Although it is unavailable for sale, MIRA is currently in the final phases of a clinical trial under an Investigational Device Exemption to accelerate U.S. Food and Drug Administration (FDA) market approval. Its size makes it appealing to surgeons and hospital staff and is ideal for use within a long-duration space mission’s tight space and mass requirements.

Once aboard, MIRA will operate inside box sized experiment locker and perform activities that simulate those real-life activities in surgery, such as cutting simulated tissue and manipulating small objects.(2)

Although it is unavailable for sale, MIRA is currently in the final phases of a clinical trial under an Investigational Device Exemption to accelerate U.S. Food and Drug Administration (FDA) market approval. Its size makes it appealing to surgeons and hospital staff and is ideal for use within a long-duration space mission’s tight space and mass requirements. Once aboard, MIRA will operate inside box sized experiment locker and perform activities that simulate those real-life activities in surgery, such as cutting simulated tissue and manipulating small objects.(2)

Researchers Aim to Use AI to Predict Rare Diseases

Researchers from the Perelman School of Medicine at the University of Pennsylvania and the University of Florida College of Medicine will explore new AI protocols with the help of the National Institutes of Health (NIH). The researchers announced their project to develop a set of machine learning-powered algorithms to determine patients at risk of five different classes of vasculitis and two different types of spondyloarthritis. These predictions, derived from data already available in patients’ electronic health records, could significantly improve the probability of patients being diagnosed sooner. The design efforts for this prediction method, called “PANDA: Predictive Analytics via Networked Distributed Algorithms for multi-system diseases, include a national database including information from different health systems, adding up to more than 27 million patients will be used to complete the algorithms. Once built, the investigators will test each algorithm’s predictive capacity across 10-plus health systems. Following these trials, the methods created by the team will be shared and available to be applied to other diseases. As its name suggests, machine learning algorithms are designed to “continue to learn” and refine themselves as they’re used and fed with more data; PANDA may continuously refine itself and become more helpful as time passes.(3)

AI and heart rate variability work on sepsis predictive models

TIIM Healthcare, an AI health technology company, and Duke-NUS medical school, both in Singapore, have received an exclusive license to commercialize a novel technology to identify patients at risk of dying from sepsis. Ai-TRIAGE software incorporates heart rate variability and another specific variable to identify patients at risk of septic adverse. The Duke-NUS technology adopts a new scoring system, which uses HRV (heart rate variability), HRnV (heart rate n-variability), vital signs, and quick sequential organ failure assessment to predict in-hospital mortality among sepsis patients in the emergency area. The solution does not require blood tests, and it can deliver risk review results as fast as 10 minutes, making it possible to be used for continuous monitoring of mortality risk among sepsis patients. The technology was developed using data obtained from about 340 sepsis patients at Singapore General Hospital’s emergency department.

Based on a study published in a peer-reviewed journal, its predictive model outperformed existing sepsis risk-scoring models. Meanwhile, TIIM Healthcare plans to integrate the novel technology into its platform to help augment clinicians’ accuracy and analytical capabilities to triage septic patients.(4)

The solution does not require blood tests, and it can deliver risk review results as fast as 10 minutes, making it possible to be used for continuous monitoring of mortality risk among sepsis patients. The technology was developed using data obtained from about 340 sepsis patients at Singapore General Hospital’s emergency department. Based on a study published in a peer-reviewed journal, its predictive model outperformed existing sepsis risk-scoring models. Meanwhile, TIIM Healthcare plans to integrate the novel technology into its platform to help augment clinicians’ accuracy and analytical capabilities to triage septic patients.(4)

Scientists train an AI model to predict breast cancer risk preventing unnecessary biopsies

Krzysztof J. Geras, an assistant professor, and his research team in the Department of Radiology at the NYU Grossman School of Medicine aim to improve breast cancer diagnostics with a new AI model. The researchers are developing AI models from large, well-annotated datasets that could be key in refining the sensitivity of these images and reducing unnecessary biopsies. The dataset consisted of bilateral DCE-MRI (Dynamic contrast-enhanced magnetic resonance imaging) studies from patients categorized into four groups: high-risk screening, preoperative planning, routine surveillance, or follow-up with a suspicious finding. They trained a deep neural network with 3D spatiotemporal features using the preciseness of the NVIDIA Apex open-source library. The model´s performance was validated on 3,936 MRIs from NYU Langone Health and three national and international databases. The team found no special statistical significance in the results between the experts and the AI system. Averaging the AI and radiologist predictions together increased overall accuracy by at least 5%, suggesting a hybrid approach may be the most beneficial. The model was also accurate among patients with various cancer subtypes, even among less common malignancies. As the guidelines reference mention, all suspicious lesions classified with category BI-RADS 4 are recommended for biopsy, which leads to a substantial number of false positives. The AI model predictions can help avoid benign biopsies in up to 20% of all BI-RADS categories. The model output can also be combined with a clinician’s or patient’s preference to decide whether to seek a biopsy behind a suspicious finding.(5)

The technology uses heart rate variability measurements to assess the severity of sepsis in ED patients

The GI Genius™ is the first-to-market, computer-aided polyp detection system powered by artificial intelligence (AI). This new software module acts like the physicians’ second set of eyes. It continuously monitors and analyzes the images during a colonoscopy and highlights suspicious polyps with a visual marker in real-time, alerting the physician to areas where it thinks a closer look is warranted. It finds them 99.7% of the time, according to data by Medtronic, which manufactures the module. According to the last analysis of the data by Medtronic, they showed that the technology reduced the number of missed colorectal polyps in a standard colonoscopy by up to 50%.(6)

According to the last analysis of the data by Medtronic, they showed that the technology reduced the number of missed colorectal polyps in a standard colonoscopy by up to 50%.(6)

Scientists are using AI to dream up revolutionary new proteins

MFDS (South Korea Ministry of Food and Drug Safety) regulators authorized the first-ever use of a completed COVID-19 vaccine from a new protein designed by The EMBL’s European Bioinformatics Institute. The latter is an Intergovernmental Organization that, as part of the European Molecular Biology Laboratory family, focuses on analysis and services in bioinformatics. They used a new vaccine from a novel protein designed by humans. The vaccine was developed on a globular protein; these advances were made more accessible and faster due to AI technology and specific projects such as the AlphaFold system by DeepMind. This power tool had predicted structures for every protein known to science. Most efforts focus on tools that can help make essential proteins shaped unlike anything in nature, without much focus on what these molecules can do. But researchers and a growing number of companies applying AI to protein design would like to design proteins that can do valuable things, from cleaning up toxic waste to treating diseases.(7)

This power tool had predicted structures for every protein known to science. Most efforts focus on tools that can help make essential proteins shaped unlike anything in nature, without much focus on what these molecules can do. But researchers and a growing number of companies applying AI to protein design would like to design proteins that can do valuable things, from cleaning up toxic waste to treating diseases.(7)

References