Potential for GPT Technology to Optimize Future Clinical Decision-Making Using Retrieval-Augmented Generation.

Advancements in artificial intelligence (AI) provide many helpful tools for healthcare, one of which includes AI chatbots that use natural language processing to create humanlike, conversational dialog. These chatbots have general cognitive skills and are able to engage with clinicians and patients to discuss patients' health conditions and what they may be at risk for. While chatbot engines have access to a wide range of medical texts and research papers, they currently provide high-level, generic responses and are limited in their ability to provide diagnostic guidance and clinical advice to patients on an individual level. The essay discusses the use of retrieval-augmented generation (RAG), which can be used to improve the specificity of user-entered prompts and thereby enhance the detail in AI chatbot responses. By embedding more recent clinical data and trusted medical sources, such as clinical guidelines, into the chatbot models, AI chatbots can provide more patient-specific guidance, faster diagnoses and treatment recommendations, and greater improvement of patient outcomes.

GPT Technology to Help Address Longstanding Barriers to Care in Free Medical Clinics.

The implementation of technology in healthcare has revolutionized patient-centered decision making by providing contextualized information about a patient's healthcare journey, leading to increased efficiency (Keyworth et al. in BMC Med Inform Decis Mak 18:93, 2018, https://doi.org/10.1186/s12911-018-0661-3 ). Artificial intelligence has been integrated within Electronic Health Records (EHR) to prompt screenings or diagnostic tests based on a patient's holistic health profile. While larger hospitals have already widely adopted these technologies, free clinics hold lower utilization of these advanced capability EHRs. The patient population at a free clinic faces a multitude of factors that limits their access to comprehensive care, thus requiring necessary efforts and measures to close the gap in healthcare disparities. Emerging Artificial Intelligence (AI) technology, such as OpenAI's ChatGPT, GPT-4, and other large language models (LLMs) have remarkable potential to improve patient care outcomes, promote health equity, and enhance comprehensive and holistic care in resource-limited settings. This paper aims to identify areas in which integrating these LLM AI advancements into free clinics operations can optimize and streamline healthcare delivery to underserved patient populations. This paper also identifies areas of improvements in GPT that are necessary to deliver those services.

Overexpression of GREM1 Improves the Survival Capacity of Aged Cardiac Mesenchymal Progenitor Cells via Upregulation of the ERK/NRF2-Associated Antioxidant Signal Pathway.

Ischemic heart disease is the leading cause of mortality in the United States. Progenitor cell therapy can restore myocardial structure and function. However, its efficacy is severely limited by cell aging and senescence. Gremlin-1 (GREM1), a member of the bone morphogenetic protein antagonist family, has been implicated in cell proliferation and survival. However, GREM1's role in cell aging and senescence has never been investigated in human cardiac mesenchymal progenitor cells (hMPCs). Therefore, this study assessed the hypothesis that overexpression of GREM1 rejuvenates the cardiac regenerative potential of aging hMPCs to a youthful stage and therefore allows better capacity for myocardial repair. We recently reported that a subpopulation of hMPCs with low mitochondrial membrane potential can be sorted from right atrial appendage-derived cells in patients with cardiomyopathy and exhibit cardiac reparative capacity in a mouse model of myocardial infarction. In this study, lentiviral particles were used to overexpress GREM1 in these hMPCs. Protein and mRNA expression were assessed through Western blot and RT-qPCR. FACS analysis for Annexin V/PI staining and lactate dehydrogenase assay were used to assess cell survival. It was observed that cell aging and cell senescence led to a decrease in GREM1 expression. In addition, overexpression of GREM1 led to a decrease in expression of senescence genes. Overexpression of GREM1 led to no significant change in cell proliferation. However, GREM1 appeared to have an anti-apoptotic effect, with an increase in survival and decrease in cytotoxicity evident in GREM1-overexpressing hMPCs. Overexpressing GREM1 also induced cytoprotective properties by decreasing reactive oxidative species and mitochondrial membrane potential. This result was associated with increased expression of antioxidant proteins, such as SOD1 and catalase, and activation of the ERK/NRF2 survival signal pathway. Inhibition of ERK led to a decrease in GREM1-mediated rejuvenation in terms of cell survival, which suggests that an ERK-dependent pathway may be involved. Taken altogether, these results indicate that overexpression of GREM1 can allow aging hMPCs to adopt a more robust phenotype with improved survival capacity, which is associated with an activated ERK/NRF2 antioxidant signal pathway.

MG53 suppresses NF-κB activation to mitigate age-related heart failure.

Aging is associated with chronic oxidative stress and inflammation that affect tissue repair and regeneration capacity. MG53 is a TRIM family protein that facilitates repair of cell membrane injury in a redox-dependent manner. Here, we demonstrate that the expression of MG53 was reduced in failing human hearts and aged mouse hearts, concomitant with elevated NF-κB activation. We evaluated the safety and efficacy of longitudinal, systemic administration of recombinant human MG53 (rhMG53) protein in aged mice. Echocardiography and pressure-volume loop measurements revealed beneficial effects of rhMG53 treatment in improving heart function of aged mice. Biochemical and histological studies demonstrated that the cardioprotective effects of rhMG53 are linked to suppression of NF-κB-mediated inflammation, reducing apoptotic cell death and oxidative stress in the aged heart. Repetitive administration of rhMG53 in aged mice did not have adverse effects on major vital organ functions. These findings support the therapeutic value of rhMG53 in treating age-related decline in cardiac function.

Calculation of physiologic dead space: comparison of ventilator volumetric capnography to measurements by metabolic analyzer and volumetric CO2 monitor.

BACKGROUND: Calculation of physiologic dead space (dead space divided by tidal volume [VD/VT]) using the Enghoff modification of the Bohr equation requires measurement of the partial pressure of mean expired CO2 (PECO2) by exhaled gas collection and analysis, use of a metabolic analyzer, or use of a volumetric CO2 monitor. The Dräger XL ventilator is equipped with integrated volumetric CO2 monitoring and calculates minute CO2 production (VCO2). We calculated PECO2 and VD/VT from ventilator derived volumetric CO2 measurements of VCO2 and compared them to metabolic analyzer and volumetric CO2 monitor measurements. METHODS: A total of 67 measurements in 36 subjects recovering from acute lung injury or ARDS were compared. Thirty-one ventilator derived measurements were compared to measurements using 3 different metabolic analyzers, and 36 ventilator derived measurements were compared to measurements from a volumetric CO2 monitor. RESULTS: There was a strong agreement between ventilator derived measurements and metabolic analyzer or volumetric CO2 monitor measurements of PECO2 and VD/VT. The correlations, bias, and precision between the ventilator and metabolic analyzer measurements for PECO2 were r = 0.97, r(2) = 0.93 (P < .001), bias -1.04 mm Hg, and precision ± 1.47 mm Hg. For VD/VT the correlations were r = 0.95 and r(2) = 0.91 (P < .001), and the bias and precision were 0.02 ± 0.04. The correlations between the ventilator and the volumetric CO2 monitor for PECO2 were r = 0.96 and r(2) = 0.92 (P < .001), and the bias and precision were -0.19 ± 1.58 mm Hg. The correlations between the ventilator and the volumetric CO2 monitor for VD/VT were r = 0.97 and r(2) = 0.95 (P < .001), and the bias and precision were 0.01 ± 0.03. CONCLUSIONS: PECO2, and therefore VD/VT, can be accurately calculated directly from the Dräger XL ventilator volumetric capnography measurements without use of a metabolic analyzer or volumetric CO2 monitor.

Identification of gastroenteric viruses by electron microscopy using higher order spectral features.

BACKGROUND: Many paediatric illnesses are caused by viral agents, for example, acute gastroenteritis. Electron microscopy can provide images of viral particles and can be used to identify the agents. OBJECTIVES: The use of electron microscopy as a diagnostic tool is limited by the need for high level of expertise in interpreting these images and the time required. A semi-automated method is proposed in this paper. STUDY DESIGN: The method is based on bispectal features that capture contour and texture information while providing robustness to shift, rotation, changes in size and noise. The magnification or true size of the viral particles need not be known precisely, but if available can be used additionally for improved classification. Viral particles from one or more images are segmented and analyzed to verify whether they belong to a particular class (such as Adenovirus, Rotavirus, etc.) or not. Two experiments were conducted-depending on the populations from which virus particle images were collected for training and testing, respectively. In the first, disjoint subsets from a pooled population of virus particles obtained from several images were used. In the second, separate populations from separate images were used. The performance of the method on viruses of similar size was separately evaluated using Astrovirus, HAV and Poliovirus. A Gaussian Mixture Model was used for the probability density of the features. A threshold on the log-likelihood is varied to study false alarm and false rejection trade-off. Features from many particles and/or likelihoods from independent tests are averaged to yield better performance. RESULTS: An equal error rate (EER) of 2% is obtained for verification of Rotavirus (tested against three other viruses) when features from 15 viral particle images are averaged. It drops further to less than 0.2% when scores from two tests are averaged to make a decision. For verification of Astrovirus (tested against two others of the same size) the EER was less than 2% when 20 particles and two tests were used. CONCLUSION: Bispectral features and Gaussian mixture modelling of their probability density are shown to be effective in identifying viruses from electron microscope images. With the use of digital imaging in electron microscopes, this method can be fully automated.