The transformative impact of large language models on medical writing and publishing: current applications, challenges and future directions.

Large language models (LLMs) are rapidly transforming medical writing and publishing. This review article focuses on experimental evidence to provide a comprehensive overview of the current applications, challenges, and future implications of LLMs in various stages of academic research and publishing process. Global surveys reveal a high prevalence of LLM usage in scientific writing, with both potential benefits and challenges associated with its adoption. LLMs have been successfully applied in literature search, research design, writing assistance, quality assessment, citation generation, and data analysis. LLMs have also been used in peer review and publication processes, including manuscript screening, generating review comments, and identifying potential biases. To ensure the integrity and quality of scholarly work in the era of LLM-assisted research, responsible artificial intelligence (AI) use is crucial. Researchers should prioritize verifying the accuracy and reliability of AI-generated content, maintain transparency in the use of LLMs, and develop collaborative human-AI workflows. Reviewers should focus on higher-order reviewing skills and be aware of the potential use of LLMs in manuscripts. Editorial offices should develop clear policies and guidelines on AI use and foster open dialogue within the academic community. Future directions include addressing the limitations and biases of current LLMs, exploring innovative applications, and continuously updating policies and practices in response to technological advancements. Collaborative efforts among stakeholders are necessary to harness the transformative potential of LLMs while maintaining the integrity of medical writing and publishing.

Drug-Drug Interactions between COVID-19 and Tuberculosis Medications: A Comprehensive Review of CYP450 and Transporter-Mediated Effects.

Most medications undergo metabolism and elimination via CYP450 enzymes, while uptake and efflux transporters play vital roles in drug elimination from various organs. Interactions often occur when multiple drugs share CYP450-transporter-mediated metabolic pathways, necessitating a unique clinical care strategy to address the diverse types of CYP450 and transporter-mediated drug-drug interactions (DDI). The primary focus of this review is to record relevant mechanisms regarding DDI between COVID-19 and tuberculosis (TB) treatments, specifically through the influence of CYP450 enzymes and transporters on drug absorption, distribution, metabolism, elimination, and pharmacokinetics. This understanding empowers clinicians to prevent subtherapeutic and supratherapeutic drug levels of COVID medications when co-administered with TB drugs, thereby mitigating potential challenges and ensuring optimal treatment outcomes. A comprehensive analysis is presented, encompassing various illustrative instances of TB drugs that may impact COVID-19 clinical behavior, and vice versa. This review aims to provide valuable insights to healthcare providers, facilitating informed decision-making and enhancing patient safety while managing co-infections. Ultimately, this study contributes to the body of knowledge necessary to optimize therapeutic approaches and improve patient outcomes in the face of the growing challenges posed by infectious diseases.

Data science through natural language with ChatGPT's Code Interpreter.

Large language models (LLMs) have emerged as a powerful tool for biomedical researchers, demonstrating remarkable capabilities in understanding and generating human-like text. ChatGPT with its Code Interpreter functionality, an LLM connected with the ability to write and execute code, streamlines data analysis workflows by enabling natural language interactions. Using materials from a previously published tutorial, similar analyses can be performed through conversational interactions with the chatbot, covering data loading and exploration, model development and comparison, permutation feature importance, partial dependence plots, and additional analyses and recommendations. The findings highlight the significant potential of LLMs in assisting researchers with data analysis tasks, allowing them to focus on higher-level aspects of their work. However, there are limitations and potential concerns associated with the use of LLMs, such as the importance of critical thinking, privacy, security, and equitable access to these tools. As LLMs continue to improve and integrate with available tools, data science may experience a transformation similar to the shift from manual to automatic transmission in driving. The advancements in LLMs call for considering the future directions of data science and its education, ensuring that the benefits of these powerful tools are utilized with proper human supervision and responsibility.

Population pharmacokinetic model of rifampicin for personalized tuberculosis pharmacotherapy: Effects of SLCO1B1 polymorphisms on drug exposure.

BACKGROUND: Rifampicin (RIF) exhibits high pharmacokinetic (PK) variability among individuals; a low plasma concentration might result in unfavorable treatment outcomes and drug resistance. This study evaluated the contributions of non- and genetic factors to the interindividual variability of RIF exposure, then suggested initial doses for patients with different weight bands. METHODS: This multicenter prospective cohort study in Korea analyzed demographic and clinical data, the solute carrier organic anion transporter family member 1B1 (SLCO1B1) genotypes, and RIF concentrations. Population PK modeling and simulations were conducted using nonlinear mixed-effect modeling. RESULTS: In total, 879 tuberculosis (TB) patients were divided into a training dataset (510 patients) and a test dataset (359 patients). A one-compartment model with allometric scaling for effect of body size best described the RIF PKs. The apparent clearance (CL/F) was 16.6% higher among patients in the SLCO1B1 rs4149056 wild-type group than among patients in variant group, significantly decreasing RIF exposure in the wild-type group. The developed model showed better predictive performance compared with previously reported models. We also suggested that patients with body weights of <40 kg, 40-55 kg, 55-70 kg, and >70 kg patients receive RIF doses of 450, 600, 750, and 1050 mg/day, respectively. CONCLUSIONS: Total body weight and SLCO1B1 rs4149056 genotypes were the most significant covariates that affected RIF CL/F variability in Korean TB patients. We suggest initial doses of RIF based on World Health Organization weight-band classifications. The model may be implemented in treatment monitoring for TB patients.

Alterations of lipid-related genes during anti-tuberculosis treatment: insights into host immune responses and potential transcriptional biomarkers.

Background: The optimal diagnosis and treatment of tuberculosis (TB) are challenging due to underdiagnosis and inadequate treatment monitoring. Lipid-related genes are crucial components of the host immune response in TB. However, their dynamic expression and potential usefulness for monitoring response to anti-TB treatment are unclear. Methodology: In the present study, we used a targeted, knowledge-based approach to investigate the expression of lipid-related genes during anti-TB treatment and their potential use as biomarkers of treatment response. Results and discussion: The expression levels of 10 genes (ARPC5, ACSL4, PLD4, LIPA, CHMP2B, RAB5A, GABARAPL2, PLA2G4A, MBOAT2, and MBOAT1) were significantly altered during standard anti-TB treatment. We evaluated the potential usefulness of this 10-lipid-gene signature for TB diagnosis and treatment monitoring in various clinical scenarios across multiple populations. We also compared this signature with other transcriptomic signatures. The 10-lipid-gene signature could distinguish patients with TB from those with latent tuberculosis infection and non-TB controls (area under the receiver operating characteristic curve > 0.7 for most cases); it could also be useful for monitoring response to anti-TB treatment. Although the performance of the new signature was not better than that of previous signatures (i.e., RISK6, Sambarey10, Long10), our results suggest the usefulness of metabolism-centric biomarkers. Conclusions: Lipid-related genes play significant roles in TB pathophysiology and host immune responses. Furthermore, transcriptomic signatures related to the immune response and lipid-related gene may be useful for TB diagnosis and treatment monitoring.

Transforming clinical trials: the emerging roles of large language models.

Clinical trials are essential for medical research, but they often face challenges in matching patients to trials and planning. Large language models (LLMs) offer a promising solution, signaling a transformative shift in the field of clinical trials. This review explores the multifaceted applications of LLMs within clinical trials, focusing on five main areas expected to be implemented in the near future: enhancing patient-trial matching, streamlining clinical trial planning, analyzing free text narratives for coding and classification, assisting in technical writing tasks, and providing cognizant consent via LLM-powered chatbots. While the application of LLMs is promising, it poses challenges such as accuracy validation and legal concerns. The convergence of LLMs with clinical trials has the potential to revolutionize the efficiency of clinical trials, paving the way for innovative methodologies and enhancing patient engagement. However, this development requires careful consideration and investment to overcome potential hurdles.

Development of a population pharmacokinetic model of pyrazinamide to guide personalized therapy: impacts of geriatric and diabetes mellitus on clearance.

Objectives: This study was performed to develop a population pharmacokinetic model of pyrazinamide for Korean tuberculosis (TB) patients and to explore and identify the influence of demographic and clinical factors, especially geriatric diabetes mellitus (DM), on the pharmacokinetics (PK) of pyrazinamide (PZA). Methods: PZA concentrations at random post-dose points, demographic characteristics, and clinical information were collected in a multicenter prospective TB cohort study from 18 hospitals in Korea. Data obtained from 610 TB patients were divided into training and test datasets at a 4:1 ratio. A population PK model was developed using a nonlinear mixed-effects method. Results: A one-compartment model with allometric scaling for body size effect adequately described the PK of PZA. Geriatric patients with DM (age >70 years) were identified as a significant covariate, increasing the apparent clearance of PZA by 30% (geriatric patients with DM: 5.73 L/h; others: 4.50 L/h), thereby decreasing the area under the concentration-time curve from 0 to 24 h by a similar degree compared with other patients (geriatric patients with DM: 99.87 µg h/mL; others: 132.3 µg h/mL). Our model was externally evaluated using the test set and provided better predictive performance compared with the previously published model. Conclusion: The established population PK model sufficiently described the PK of PZA in Korean TB patients. Our model will be useful in therapeutic drug monitoring to provide dose optimization of PZA, particularly for geriatric patients with DM and TB.

Impact of Criterion Versus Norm-Referenced Assessment on the Quality of Life in Korean Medical Students.

BACKGROUND: Medical students are known to be subjected to immense stress under competitive curricula and have a high risk of depression, burnout, anxiety and sleep disorders. There is a global trend of switching from norm-referenced assessment (NRA) to criterion-referenced assessment (CRA), and these changes may have influenced the quality of life (QOL), sleep phase, sleep quality, stress, burnout, and depression of the medical students. We hypothesized that there is a significant difference of QOL between CRA and NRA and that sleep, stress, burnout, and depression are the main contributors. METHODS: By administering an online survey regarding QOL and its contributors to Korean medical students, 365 responses from 10 medical schools were recorded. To clarify the complex relationship between the multiple factors in play, we applied nonlinear machine learning algorithms and utilized causal structure learning techniques on the survey data. RESULTS: Students with CRA had lower scores in stress (68.16 ± 11.29, 76.03 ± 12.38, P < 0.001), burnout (48.09 ± 11.23, 55.93 ± 13.07, P < 0.001), depression (12.77 ± 9.82, 16.44 ± 11.27, P = 0.003) and higher scores in QOL (95.79 ± 16.20, 89.65 ± 16.28, P < 0.001) compared with students with NRA. Multiple linear regression, permutation importance of the random forest model and the causal structure model showed that depression, stress and burnout are the most influential factors of QOL of medical students. CONCLUSION: Medical students from schools that use CRA showed higher QOL scores, as well as lower burnout, stress and depression when compared with students from schools that use NRA. These results may be used as a basis for granting justification for the transition to CRA.

Recommendation of pharmacokinetics/pharmacodynamics target of ethambutol to suppress tuberculosis resistance: A population pharmacokinetics study on a large prospective cohort.

BACKGROUND: The ability of ethambutol (EMB) to suppress bacterial resistance has been demonstrated in a time-dependent manner. Through the development of a population pharmacokinetics (PK) model, this study aimed to suggest the PK/pharmacodynamics (PD) target and identify the significant covariates that influence interindividual variability (IIV) in the PK of EMB. METHODS: In total, 837 patients from 20 medical centres across Korea were enrolled in this study. The non-linear mixed-effect method was used to establish and validate the population PK model. RESULTS: A two-compartment model with transit compartment absorption was sufficient to describe the PK of EMB. Body weight and renal function were identified as significant covariates that affect IIV of the apparent clearance (CL/F) of EMB. Patients with moderate renal function showed 35% and 55% lower CL/F (CL/F 89.9 L/h) compared with those with mild and normal renal function, respectively. All the renal function groups with simulated doses ranging from 800 to 1200 mg achieved area under the curve over minimum inhibitory concentration (MIC) >119, and maintained T>MIC for >23 h for MIC of 0.5 µg/mL. Based on our simulation result, it is suggested that doses of 800, 1000, and 1200 mg should obtain the T>MIC target of 4, 6, and 8 h, respectively. This model was validated internally and externally. CONCLUSION: This study provides insight into the PK/PD indexes of EMB for three different renal function groups and T>MIC targets for different doses. The results could be used to provide optimal-dose suggestions for EMB.

CYP2C19 Contributes to THP-1-Cell-Derived M2 Macrophage Polarization by Producing 11,12- and 14,15-Epoxyeicosatrienoic Acid, Agonists of the PPARγ Receptor.

Although the functional roles of M1 and M2 macrophages in the immune response and drug resistance are important, the expression and role of cytochrome P450s (CYPs) in these cells remain largely unknown. Differential expression of the 12 most common CYPs (CYP1A1, 1A2, 1B1, 2B6, 2C8, 2C9, 2C19, 2D6, 2E1, 2J2, 3A4, and 3A5) were screened in THP-1-cell-derived M1 and M2 macrophages using reverse transcription PCR. CYP2C19 was highly expressed in THP-1-cell-derived M2 macrophages, but it was negligibly expressed in THP-1-cell-derived M1 macrophages at the mRNA and protein levels as analyzed by reverse transcription quantitative PCR and Western blot, respectively. CYP2C19 enzyme activity was also very high in THP-1-cell-derived M2 compared to M1 macrophages (> 99%, p < 0.01), which was verified using inhibitors of CYP2C19 activity. Endogenous levels of the CYP2C19 metabolites 11,12-epoxyeicosatrienoic acid (11,12-EET) and 14,15-EET were reduced by 40% and 50% in cells treated with the CYP2C19 inhibitor and by 50% and 60% in the culture medium, respectively. Both 11,12-EET and 14,15-EET were identified as PPARγ agonists in an in vitro assay. When THP-1-cell-derived M2 cells were treated with CYP2C19 inhibitors, 11,12- and 14,15-EETs were significantly reduced, and in parallel with the reduction of these CYP2C19 metabolites, the expression of M2 cell marker genes was also significantly decreased (p < 0.01). Therefore, it was suggested that CYP2C19 may contribute to M2 cell polarization by producing PPARγ agonists. Further studies are needed to understand the endogenous role of CYP2C19 in M2 macrophages with respect to immunologic function and cell polarization.

Multimodal plasma metabolomics and lipidomics in elucidating metabolic perturbations in tuberculosis patients with concurrent type 2 diabetes.

Type 2 diabetes mellitus (DM) poses a major burden for the treatment and control of tuberculosis (TB). Characterization of the underlying metabolic perturbations in DM patients with TB infection would yield insights into the pathophysiology of TB-DM, thus potentially leading to improvements in TB treatment. In this study, a multimodal metabolomics and lipidomics workflow was applied to investigate plasma metabolic profiles of patients with TB and TB-DM. Significantly different biological processes and biomarkers in TB-DM vs. TB were identified using a data-driven, knowledge-based framework. Changes in metabolic and signaling pathways related to carbohydrate and amino acid metabolism were mainly captured by amide HILIC column metabolomics analysis, while perturbations in lipid metabolism were identified by the C18 metabolomics and lipidomics analysis. Compared to TB, TB-DM exhibited elevated levels of bile acids and molecules related to carbohydrate metabolism, as well as the depletion of glutamine, retinol, lysophosphatidylcholine, and phosphatidylcholine. Moreover, arachidonic acid metabolism was determined as a potentially important factor in the interaction between TB and DM pathophysiology. In a correlation network of the significantly altered molecules, among the central nodes, chenodeoxycholic acid was robustly associated with TB and DM. Fatty acid (22:4) was a component of all significant modules. In conclusion, the integration of multimodal metabolomics and lipidomics provides a thorough picture of the metabolic changes associated with TB-DM. The results obtained from this comprehensive profiling of TB patients with DM advance the current understanding of DM comorbidity in TB infection and contribute to the development of more effective treatment.

Development of population pharmacokinetics model and Bayesian estimation of rifampicin exposure in Indonesian patients with tuberculosis.

BACKGROUND: Interindividual variability in the pharmacokinetics (PK) of anti-tuberculosis (TB) drugs is the leading cause of treatment failure. Herein, we evaluated the influence of demographic, clinical, and genetic factors that cause variability in RIF PK parameters in Indonesian TB patients. METHODS: In total, 210 Indonesian patients with TB (300 plasma samples) were enrolled in this study. Clinical data, solute carrier organic anion transporter family member-1B1 (SLCO1B1) haplotypes *1a, *1b, and *15, and RIF concentrations were analyzed. The population PK model was developed using a non-linear mixed effect method. RESULTS: A one-compartment model with allometric scaling adequately described the PK of RIF. Age and SLCO1B1 haplotype *15 were significantly associated with variability in apparent clearance (CL/F). For patients in their 40s, each 10-year increase in age was associated with a 10% decrease in CL/F (7.85 L/h). Patients with the SLCO1B1 haplotype *15 had a 24% lower CL/F compared to those with the wild-type. Visual predictive checks and non-parametric bootstrap analysis indicated good model performance. CONCLUSION: Age and SLCO1B1 haplotype *15 were significant covariates of RIF CL/F. Geriatric patients with haplotype *15 had significantly greater exposure to RIF. The model could optimize TB pharmacotherapy through its application in therapeutic drug monitoring (clinical trial no. NCT05280886).

Tumor Microenvironment and Genes Affecting the Prognosis of Temozolomide-Treated Glioblastoma.

Glioblastoma (GBM) is the most frequent primary brain tumor in adults and has a poor prognosis due to its resistance to Temozolomide (TMZ). However, there is limited research regarding the tumor microenvironment and genes related to the prognosis of TMZ-treated GBM patients. This study aimed to identify putative transcriptomic biomarkers with predictive value in patients with GBM who were treated with TMZ. Publicly available datasets from The Cancer Genome Atlas and Gene Expression Omnibus were analyzed using CIBERSORTx and Weighted Gene Co-expression Network Analysis (WGCNA) to obtain types of highly expressed cell types and gene clusters. Differentially Expressed Genes analysis was performed and was intersected with the WGCNA results to obtain a candidate gene list. Cox proportional-hazard survival analysis was performed to acquire genes related to the prognosis of TMZ-treated GBM patients. Inflammatory microglial cells, dendritic cells, myeloid cells, and glioma stem cells were highly expressed in GBM tissue, and ACP7, EPPK1, PCDHA8, RHOD, DRC1, ZIC3, and PRLR were significantly associated with survival. While the listed genes have been previously reported to be related to glioblastoma or other types of cancer, ACP7 was identified as a novel gene related to the prognosis of GBM. These findings may have potential implications for developing a diagnostic tool to predict GBM resistance and optimize treatment decisions.

Building and analyzing machine learning-based warfarin dose prediction models using scikit-learn.

For personalized drug dosing, prediction models may be utilized to overcome the inter-individual variability. Multiple linear regression has been used as a conventional method to model the relationship between patient features and optimal drug dose. However, linear regression cannot capture non-linear relationships and may be adversely affected by non-normal distribution and collinearity of data. To overcome this hurdle, machine learning models have been extensively adapted in drug dose prediction. In this tutorial, random forest and neural network models will be trained in tandem with a multiple linear regression model on the International Warfarin Pharmacogenetics Consortium dataset using the scikit-learn python library. Subsequent model analyses including performance comparison, permutation feature importance computation and partial dependence plotting will be demonstrated. The basic methods of model training and analysis discussed in this article may be implemented in drug dose-related studies.

Comparison of multivariate linear regression and a machine learning algorithm developed for prediction of precision warfarin dosing in a Korean population.

BACKGROUND: Personalized warfarin dosing is influenced by various factors including genetic and non-genetic factors. Multiple linear regression (LR) is known as a conventional method to develop predictive models. Recently, machine learning approaches have been extensively implemented for warfarin dosing due to the hypothesis of non-linear association between covariates and stable warfarin dose. OBJECTIVE: To extend the multiple linear regression algorithm for personalized warfarin dosing in a Korean population and compare with a machine learning--based algorithm. METHOD: From this cohort study, we collected information on 650 patients taking warfarin who achieved steady state including demographic information, indications, comorbidities, comedications, habits, and genetic factors. The dataset was randomly split into training set (90%) and test set (10%). The LR and machine learning (gradient boosting machine [GBM]) models were developed on the training set and were evaluated on the test set. RESULT: LR and GBM models were comparable in terms of accuracy of ideal dose (75.38% and 73.85%), correlation (0.77 and 0.73), mean absolute error (0.58 mg/day and 0.64 mg/day), and root mean square error (0.82 mg/day and 0.9 mg/day), respectively. VKORC1 genotype, CYP2C9 genotype, age, and weight were the highest contributors and could obtain 80% of maximum performance in both models. CONCLUSION: This study shows that our LR and GMB models are satisfactory to predict warfarin dose in our dataset. Both models showed similar performance and feature contribution characteristics. LR may be the appropriate model due to its simplicity and interpretability.

Pharmacokinetics of fixed-dose combination of atorvastatin and metformin compared with individual tablets.

Purpose: The aims of this study was to investigate the mutual pharmacokinetic interactions between steady-state atorvastatin and metformin and the effect of food on the fixed-dose combined (FDC) tablet of atorvastatin and metformin extended release (XR). Subjects and methods: Study 1, an open-labeled, fixed sequence, multiple-dose pharmacokinetic drug-drug interaction study, was divided into 2 parts. Atorvastatin (40 mg) or metformin (1,000 mg) XR tablets were administered once daily via mono- or co-therapy for 7 days. Plasma levels of atorvastatin and 2-OH-atorvastatin, were quantitatively determined for 36 h in part A (n=50) while metformin plasma concentration was measured up to 24 h in part B (n=16) after the last dosing. Study 2, a randomized, open-labeled, single-dose, two-treatment, two-period, two-sequence crossover study, involved 27 healthy subjects to investigate the impact of food intake on the pharmacokinetics of a combined atorvastatin/metformin XR 20/500 mg (CJ-30056 20/500 mg) tablet. Results: After multiple doses of mono- or co-therapy of atorvastatin (40 mg) and metformin (1,000 mg) XR, the 90% confidence intervals (CIs) of the geometric mean ratios (GMRs) for the peak plasma concentration at steady state (Cmax,ss) and area under the plasma concentration-time curve during the dosing interval at steady state (AUCτ,ss) were 1.07 (0.94-1.22) and 1.05 (0.99-1.10) for atorvastatin, 1.06 (0.96-1.16) and 1.16 (1.10-1.21) for 2-OH-atorvastatin, and 1.00 (0.86-1.18) and 0.99 (0.87-1.13) for metformin, respectively. Food delayed time to reach maximum concentration (tmax), decreased atorvastatin Cmax by 32% with a GMR (90% CI) of 0.68 (0.59-0.78), and increased metformin AUCt by 56% with a GMR (90% CI) of 1.56 (1.43-1.69). Conclusion: No clinically relevant pharmacokinetic interaction was seen when atorvastatin was co-administered with metformin. Food appeared to change the absorption of atorvastatin and metformin from an FDC formulation. These alterations were in accordance with those described with the single reference drugs when ingested with food.

A new population pharmacokinetic model for vancomycin in patients with variable renal function: Therapeutic drug monitoring based on extended covariate model using CKD-EPI estimation.

WHAT IS KNOWN AND OBJECTIVE: Although patients may have received vancomycin therapy with therapeutic drug monitoring (TDM), those treated with high-strength and long-term vancomycin therapy might have unstable and time-varying renal function. The methods used to estimate renal function should not be considered interchangeable with pharmacokinetic (PK) modeling and model-based estimation of vancomycin pharmacokinetics. While Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) for renal function estimation has been widely integrated into clinical practice, a population PK model including CKD-EPI has not been established. The study was aimed at developing a new population PK model for optimal vancomycin prediction in patients with time-varying and variable renal function to evaluate the interchangeability of estimation methods. METHODS: The most suitable population PK model was explored and evaluated using non-linear mixed-effect modelling for the best fit of vancomycin concentrations from patients who needed to maintain high trough vancomycin concentrations of >10 mg/L or >15 mg/L. Renal function was estimated using the Cockcroft-Gault (CG), Modification of Diet in Renal Disease (MDRD) and CKD-EPI equations. NONMEM 7.4 was used to develop the population PK model. RESULTS: A total of 328 vancomycin concentrations in 99 patients were used to develop the population PK model. Vancomycin pharmacokinetics was best described by a two-compartment model. The CKD-EPI equation for vancomycin clearance was included in the final model among the estimation methods of renal function. A new covariate model, including extended covariate parameters that explain changes in renal function from the population-predicted value and individual dosing time, provided the best explanation for vancomycin pharmacokinetics among the various models tested. WHAT IS NEW AND CONCLUSION: A new extended covariate model for vancomycin using the CKD-EPI method may afford suitable dose adjustment for high-strength and long-term vancomycin therapy that results in unstable renal function.