The potential and pitfalls of using a large language model such as ChatGPT, GPT-4, or LLaMA as a clinical assistant.

OBJECTIVES: This study aims to evaluate the utility of large language models (LLMs) in healthcare, focusing on their applications in enhancing patient care through improved diagnostic, decision-making processes, and as ancillary tools for healthcare professionals. MATERIALS AND METHODS: We evaluated ChatGPT, GPT-4, and LLaMA in identifying patients with specific diseases using gold-labeled Electronic Health Records (EHRs) from the MIMIC-III database, covering three prevalent diseases-Chronic Obstructive Pulmonary Disease (COPD), Chronic Kidney Disease (CKD)-along with the rare condition, Primary Biliary Cirrhosis (PBC), and the hard-to-diagnose condition Cancer Cachexia. RESULTS: In patient identification, GPT-4 had near similar or better performance compared to the corresponding disease-specific Machine Learning models (F1-score ≥ 85%) on COPD, CKD, and PBC. GPT-4 excelled in the PBC use case, achieving a 4.23% higher F1-score compared to disease-specific "Traditional Machine Learning" models. ChatGPT and LLaMA3 demonstrated lower performance than GPT-4 across all diseases and almost all metrics. Few-shot prompts also help ChatGPT, GPT-4, and LLaMA3 achieve higher precision and specificity but lower sensitivity and Negative Predictive Value. DISCUSSION: The study highlights the potential and limitations of LLMs in healthcare. Issues with errors, explanatory limitations and ethical concerns like data privacy and model transparency suggest that these models would be supplementary tools in clinical settings. Future studies should improve training datasets and model designs for LLMs to gain better utility in healthcare. CONCLUSION: The study shows that LLMs have the potential to assist clinicians for tasks such as patient identification but false positives and false negatives must be mitigated before LLMs are adequate for real-world clinical assistance.

Engineered non-covalent π interactions as key elements for chiral recognition.

Molecular recognition and self-assembly are often mediated by intermolecular forces involving aromatic π-systems. Despite the ubiquity of such interactions in biological systems and in the design of functional materials, the elusive nature of aromatic π interaction results in that they have been seldom used as a design element for promoting challenging chemical reactions. Described here is a well-engineered catalytic system into which non-covalent π interactions are directly incorporated. Enabled by a lone pair-π interaction and a π-π stacking interaction operating collectively, efficient chiral recognition is successfully achieved in the long-pursued dihydroxylation-based kinetic resolution. Density functional theory calculations shed light on the crucial role played by the lone pair-π interaction between the carbonyl oxygen of the cinchona alkaloid ligand and the electron-deficient phthalazine π moiety of the substrate in the stereoselectivity-determining transition states. This discovery serves as a proof-of-principle example showing how the weak non-covalent π interactions, if ingeniously designed, could be a powerful guide in attaining highly enantioselective catalysis.

Simultaneous Kinetic Resolution and Asymmetric Induction within a Borrowing Hydrogen Cascade Mediated by a Single Catalyst.

The mechanistic uniqueness and versatility of borrowing hydrogen catalysis provides an opportunity to investigate the controllability of a cascade reaction, and more importantly, to realize either one or both of chiral recognition and chiral induction simultaneously. Here we report that, in a borrowing hydrogen cascade starting from racemic allylic alcohols, one of the enantiomers could be kinetically resolved, while the other enantiomer could be purposely converted to various targeted products, including α,ß-unsaturated ketones, ß-functionalized ketones and γ-functionalized alcohols. By employing a robust Ru-catalyst, both kinetic resolution and asymmetric induction were achieved with remarkable levels of efficiency and enantioselectivity. Density functional theory (DFT) calculations suggest that corresponding catalyst-substrate π-π interactions are pivotal to realize the observed stereochemical diversity.

Highly Efficient Kinetic Resolution of Aryl-Alkenyl Alcohols by Ru-Catalyzed Hydrogen Transfer.

No matter through asymmetric reduction of ketones or kinetic resolution of secondary alcohols, enantioselective synthesis of the corresponding secondary alcohols is challenging when the two groups attached to the prochiral or chiral centers are spatially or electronically similar. For examples, dialkyl (sp3 vs. sp3), diaryl (sp2 vs. sp2), and aryl-alkenyl (sp2 vs. sp2) alcohols are difficult to produce with high enantioselectivities. By exploiting our recently developed Ru-catalysts of minimal stereogenicity, we reported herein a highly efficient kinetic resolution of aryl-alkenyl alcohols through hydrogen transfer. This method enabled such versatile chiral building blocks for organic synthesis as allylic alcohols, to be readily accessed with excellent enantiomeric excesses at practically useful conversions.

Development of highly efficient platinum catalysts for hydroalkoxylation and hydroamination of unactivated alkenes.

Hydrofunctionalization, the direct addition of an X-H (e.g., X=O, N) bond across an alkene, is a desirable strategy to make heterocycles that are important structural components of naturally occurring molecules. Described here is the design and discovery of "donor-acceptor"-type platinum catalysts that are highly effective in both hydroalkoxylation and hydroamination of unactivated alkenes over a broad range of substrates under mild conditions. A number of alkene substitution patterns are accommodated, including tri-substituted, 1,1-disubstituted, (E)-disubstituted, (Z)-disubstituted and even mono-substituted double bonds. Detailed mechanistic investigations suggest a plausible pathway that includes an unexpected dissociation/re-association of the electron-deficient ligand to form an alkene-bound "donor-acceptor"-type intermediate. These mechanistic studies help understand the origins of the high reactivity exhibited by the catalytic system, and provide a foundation for the rational design of chiral catalysts towards asymmetric hydrofunctionalization reactions.

Zwitterionic nido-Carborane-Fused Phospholes.

Using readily available and easy-to-handle reagents, a novel type of nido-carborane-fused phosphole was obtained from a simple procedure. Its crystallographic structure was determined, and DFT calculations have been performed, unraveling the key influence of the zwitterionic structure on the photophysical properties.

Initial Self-Healing Temperatures of Asphalt Mastics Based on Flow Behavior Index.

Increasing temperature is a simple and convenient method to accelerate the self-healing process of bitumen. However, bitumen may not achieve the healing capability at lower temperature, and may be aged if temperature is too high. In addition, the bitumen is mixed with mineral filler and formed as asphalt mastic in asphalt concrete, so it is more accurate to study the initial self-healing from the perspective of asphalt mastic. The primary purpose of this research was to examine the initial self-healing temperature of asphalt mastic, which was determined by the flow behavior index obtained from the flow characteristics. Firstly, the texture and geometry characteristics of two fillers were analyzed, and then the initial self-healing temperature of nine types of asphalt mastic, pure bitumen (PB) and styrene-butadiene-styrene (SBS) modified bitumen were determined by the flow behavior index. Results demonstrate that the average standard deviation of gray-scale texture value of limestone filler (LF) is 21.24% lower than that of steel slag filler (SSF), showing that the steel slag filler has a better particle distribution and geometry characteristics. Also the initial self-healing temperatures of asphalt mastics with 0.2, 0.4 and 0.6 LF-PB volume ratio are 46.5 °C, 47.2 °C and 49.4 °C, which are 1.4 °C, 0.8 °C and 0.4 °C higher than that of asphalt mastics with SSF-PB, but not suitable for the evaluation of asphalt mastic contained SBS modified bitumen because of unique structure and performance of SBS.

Prevention of prostate cancer by natural product MDM2 inhibitor GS25: in vitro and in vivo activities and molecular mechanisms.

Prostate cancer remains a major health problem in the USA and worldwide. There is an urgent need to develop novel approaches to preventing primary and metastatic prostate cancer. We have identified 25-OCH3-protopanaxadiol (GS25), the most active ginsenoside that has been identified so far; it has potent activity against human cancers, including prostate cancer. However, it has not been proven if GS25 could be a safe and effective agent for cancer prevention. In this study, we used the TRAMP model and clearly demonstrated that GS25 inhibited prostate tumorigenesis and metastasis with minimal host toxicity. Mechanistically, GS25 directly bound to the RING domain of MDM2, disrupted MDM2-MDMX binding and induced MDM2 protein degradation, resulting in strong inhibition of prostate cancer cell growth and metastasis, independent of p53 and androgen receptor status. In conclusion, our in vitro and in vivo data support the potential use of GS25 in prevention of primary and metastatic prostate cancer.

Feruloylated oligosaccharides from maize bran alleviate the symptoms of diabetes in streptozotocin-induced type 2 diabetic rats.

This study investigated the therapeutic effect of feruloylated oligosaccharides (FOs) extracted from maize bran on type 2 diabetic rats and its potential mechanism. Streptozotocin (STZ) induced type 2 diabetic male rats were orally administered with different levels of FOs for 8 weeks, and ferulic acid (FA) treatment was conducted as the positive control. Among all the treatments, the oral administration of 600 mg per kg bw per d FOs showed the best therapeutic effects on the diabetic rats by significantly lowering the levels of fasting plasma glucose (FPG), fasting insulin, TG, LDL-c, aspartate transaminase, creatine kinase and lactate dehydrogenase in plasma, while increasing the level of plasma HDL-c. In addition, the intake of FOs at 600 mg per kg bw per d exhibited the best antioxidant effects in the plasma, liver, kidney and heart of the diabetic rats, and the highest inhibitory effects on the formation of AGEs and CML in the organs, which might explain the alleviating effects of FOs on abdominal aorta injury observed in the current study. FOs presented better regulation effects on FPG, plasma lipid and the protection of abdominal aorta than FA under the same administered dosage. Based on these outcomes, FOs from maize bran could be beneficial for prevention or early treatment of diabetes mellitus.

Selective Synthesis of (Z)-Diazadiphosphafulvalene from 2,2'-bis-Azaphosphindole.

The unprecedented 2,2'-bis(azaphosphindole) has been synthesized via a new route. Reaction with NaH afforded a dianion derivative 5, which is easily transformed to alkylated bis(azaphosphindole) or (Z)-P,P,N,N-cisoid diazadiphosphafulvalene. The reaction features good regioselectivity and high steroselectivity. Relatively strong fluorescence is observed with diazadiphosphafulvalenes. The X-ray crystal structure analysis showed that dianion ligand 5 is bonded to two Na atoms in a bridging cis-fashion, which allows the synthesis of diazadiphosphafulvalene in a highly stereoselective approach.