Ethical and regulatory challenges of large language models in medicine.

With the rapid growth of interest in and use of large language models (LLMs) across various industries, we are facing some crucial and profound ethical concerns, especially in the medical field. The unique technical architecture and purported emergent abilities of LLMs differentiate them substantially from other artificial intelligence (AI) models and natural language processing techniques used, necessitating a nuanced understanding of LLM ethics. In this Viewpoint, we highlight ethical concerns stemming from the perspectives of users, developers, and regulators, notably focusing on data privacy and rights of use, data provenance, intellectual property contamination, and broad applications and plasticity of LLMs. A comprehensive framework and mitigating strategies will be imperative for the responsible integration of LLMs into medical practice, ensuring alignment with ethical principles and safeguarding against potential societal risks.

Artificial intelligence, ChatGPT, and other large language models for social determinants of health: Current state and future directions.

This perspective highlights the importance of addressing social determinants of health (SDOH) in patient health outcomes and health inequity, a global problem exacerbated by the COVID-19 pandemic. We provide a broad discussion on current developments in digital health and artificial intelligence (AI), including large language models (LLMs), as transformative tools in addressing SDOH factors, offering new capabilities for disease surveillance and patient care. Simultaneously, we bring attention to challenges, such as data standardization, infrastructure limitations, digital literacy, and algorithmic bias, that could hinder equitable access to AI benefits. For LLMs, we highlight potential unique challenges and risks including environmental impact, unfair labor practices, inadvertent disinformation or "hallucinations," proliferation of bias, and infringement of copyrights. We propose the need for a multitiered approach to digital inclusion as an SDOH and the development of ethical and responsible AI practice frameworks globally and provide suggestions on bridging the gap from development to implementation of equitable AI technologies.

The next generation of healthcare ecosystem in the metaverse.

The Metaverse has gained wide attention for being the application interface for the next generation of Internet. The potential of the Metaverse is growing, as Web 3·0 development and adoption continues to advance medicine and healthcare. We define the next generation of interoperable healthcare ecosystem in the Metaverse. We examine the existing literature regarding the Metaverse, explain the technology framework to deliver an immersive experience, along with a technical comparison of legacy and novel Metaverse platforms that are publicly released and in active use. The potential applications of different features of the Metaverse, including avatar-based meetings, immersive simulations, and social interactions are examined with different roles from patients to healthcare providers and healthcare organizations. Present challenges in the development of the Metaverse healthcare ecosystem are discussed, along with potential solutions including capabilities requiring technological innovation, use cases requiring regulatory supervision, and sound governance. This proposed concept and framework of the Metaverse could potentially redefine the traditional healthcare system and enhance digital transformation in healthcare. Similar to AI technology at the beginning of this decade, real-world development and implementation of these capabilities are relatively nascent. Further pragmatic research is needed for the development of an interoperable healthcare ecosystem in the Metaverse.

A Retrospective Cohort Study Evaluating the Safety and Efficacy of Sequential versus Concurrent Intrapleural Instillation of Tissue Plasminogen Activator and DNase for Pleural Infection.

Methods: We conducted a retrospective review of patients with pleural infection requiring intrapleural therapy at two tertiary referral centres. Results: We included 84 (62.2%) and 51 (37.8%) patients who received sequential and concurrent intrapleural therapy, respectively. Patient demographics and clinical characteristics, including age, RAPID score, and percentage of pleural opacity on radiographs before intrapleural therapy, were similar in both groups. Treatment failure rates (defined by either in-hospital mortality, surgical intervention, or 30-day readmission for pleural infection) were 9.5% and 5.9% with sequential and concurrent intrapleural therapy, respectively (p = 0.534). This translates to a treatment success rate of 90.5% and 94.1% for sequential and concurrent intrapleural therapy, respectively. There was no significant difference in the decrease in percentage of pleural effusion size on chest radiographs (15.1% [IQR 6-35.7] versus 26.6% [IQR 9.9-38.7], p = 0.143) between sequential and concurrent therapy, respectively. There were also no significant differences in the rate of pleural bleeding (4.8% versus 9.8%, p = 0.298) and chest pain (13.1% versus 9.8%, p = 0.566) between sequential and concurrent therapy, respectively. Conclusion: Our study adds to the growing literature on the safety and efficacy of concurrent intrapleural therapy in pleural infection.

Artificial intelligence education: An evidence-based medicine approach for consumers, translators, and developers.

Current and future healthcare professionals are generally not trained to cope with the proliferation of artificial intelligence (AI) technology in healthcare. To design a curriculum that caters to variable baseline knowledge and skills, clinicians may be conceptualized as "consumers", "translators", or "developers". The changes required of medical education because of AI innovation are linked to those brought about by evidence-based medicine (EBM). We outline a core curriculum for AI education of future consumers, translators, and developers, emphasizing the links between AI and EBM, with suggestions for how teaching may be integrated into existing curricula. We consider the key barriers to implementation of AI in the medical curriculum: time, resources, variable interest, and knowledge retention. By improving AI literacy rates and fostering a translator- and developer-enriched workforce, innovation may be accelerated for the benefit of patients and practitioners.

A translational perspective towards clinical AI fairness.

Artificial intelligence (AI) has demonstrated the ability to extract insights from data, but the fairness of such data-driven insights remains a concern in high-stakes fields. Despite extensive developments, issues of AI fairness in clinical contexts have not been adequately addressed. A fair model is normally expected to perform equally across subgroups defined by sensitive variables (e.g., age, gender/sex, race/ethnicity, socio-economic status, etc.). Various fairness measurements have been developed to detect differences between subgroups as evidence of bias, and bias mitigation methods are designed to reduce the differences detected. This perspective of fairness, however, is misaligned with some key considerations in clinical contexts. The set of sensitive variables used in healthcare applications must be carefully examined for relevance and justified by clear clinical motivations. In addition, clinical AI fairness should closely investigate the ethical implications of fairness measurements (e.g., potential conflicts between group- and individual-level fairness) to select suitable and objective metrics. Generally defining AI fairness as "equality" is not necessarily reasonable in clinical settings, as differences may have clinical justifications and do not indicate biases. Instead, "equity" would be an appropriate objective of clinical AI fairness. Moreover, clinical feedback is essential to developing fair and well-performing AI models, and efforts should be made to actively involve clinicians in the process. The adaptation of AI fairness towards healthcare is not self-evident due to misalignments between technical developments and clinical considerations. Multidisciplinary collaboration between AI researchers, clinicians, and ethicists is necessary to bridge the gap and translate AI fairness into real-life benefits.

Therapeutic drug monitoring of meropenem and piperacillin-tazobactam in the Singapore critically ill population - A prospective, multi-center, observational study (BLAST 1).

PURPOSE: To determine percentage of patients with sub-therapeutic beta-lactam exposure in our intensive care units (ICU) and to correlate target attainment with clinical outcomes. MATERIALS AND METHODS: Multi-centre, prospective, observational study was conducted in ICUs from three hospitals in Singapore from July 2016 to May 2018. Adult patients (≥21 years) receiving meropenem or piperacillin-tazobactam were included. Four blood samples were obtained during a dosing interval to measure and determine attainment of therapeutic targets: unbound beta-lactam concentration above (i) minimum inhibitory concentration (MIC) at 40% (meropenem) or 50% (piperacillin) of dosing interval (40-50%fT > MIC) and (ii) 5 × MIC at 100% of dosing interval (100%fT > 5 × MIC). Correlation to clinical outcomes was evaluated using Cox regression. RESULTS: Beta-lactam levels were highly variable among 61 patients, with trough meropenem and piperacillin levels at 21.5 ± 16.8 mg/L and 101.6 ± 81.1 mg/L respectively. Among 85 sets of blood samples, current dosing practices were able to achieve 94% success for 40-50%fT > MIC and 44% for 100%fT > 5 × MIC. Failure to achieve 40-50%fT > MIC within 48 h was significantly associated with all-cause mortality (HR: 9.0, 95% CI: 1.8-45.0), after adjustment for APACHE II score. Achievement of 100%fT > 5 × MIC within 48 h was significantly associated with shorter length of hospital stay. CONCLUSION: Current dosing practices may be suboptimal for ICU patients. Beta-lactam TDM may be useful.

An open label randomized trial to assess the efficacy of tranexamic acid in reducing post-operative recurrence of chronic subdural haemorrhage.

Chronic subdural haemorrhage (CSDH) is a common neurosurgical entity with complex pathophysiological pathways. The generally favourable surgical outcome may be affected by its associated risks including recurrence rates. We performed a prospective randomized multi-center clinical trial comparing the addition of tranexamic acid (TXA) to standard neurosurgical procedures for patients with symptomatic CSDH. The primary endpoint was CSDH requiring repeat surgery within 6-month post-operatively. Secondary endpoints were comparison of post-operative volumes between the treatment arms and safety evaluation of the dosing regime. 90 patients were analyzed with 49 patients in the observation arm and 41 patients in the TXA arm. The observation arm had five (10.2%) recurrences compared to two (4.8%, p = 0.221) in the TXA arm. Patients in the TXA arm demonstrated a greater reduction of their CSDH volume at 6 weeks follow up (36.6%) compared to the observation arm (23.3%, p = 0.6648). There were no reportable serious adverse events recorded in the observation arm, compared to 4 (9.8%) patients in the TXA arm. The addition of TXA treatment to standard surgical drainage of CSH did not significantly reduce symptomatic post-operative recurrence. Patients in the TXA arm had a delay in the CSDH recurrence with a comparative reduction of residual hematoma volume at the 6-week follow up although the effect was unsustained. Larger randomized trials with dose adjustments should be considered to investigate subgroups of patients that may benefit from this medical adjunct.

Development of stealth liposome coencapsulating doxorubicin and fluoxetine.

Stealth liposomes form an important subset of liposomes, demonstrating prolonged circulation half-life and improved safety in vivo. Caelyx® (liposomal doxorubicin; Merck & Co., Whitehouse Station, New Jersey, USA) is a successful example of the application of stealth liposomes in anticancer treatment. However, multidrug resistance (MDR) to chemotherapy still remains a critical problem, accounting for more than 90% of treatment failure in patients with advanced cancer. To circumvent MDR, fluoxetine and doxorubicin were tested in combination for synergistic activity in MCF-7 (human breast carcinoma) and MCF-7/adr (doxorubicin-resistant human breast carcinoma) cells using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) cell-viability assay. Coencapsulation of doxorubicin and fluoxetine, using an ammonium sulphate gradient, was investigated, and a factorial experiment was designed to determine the optimal drug-to-lipid (D/L) ratio for coencapsulation. Drug release from Dox-Flu-SL (stealth liposome coencapsulating doxorubicin and fluoxetine) under both in vitro and in vivo conditions was determined. In MCF-7 cells, synergism was demonstrated at specific doxorubicin:fluoxetine ratios of between 0.09 and 0.5 (molar ratio), while MCF/7/adr cells demonstrated synergism across all drug ratios. Coencapsulation of doxorubicin and fluoxetine (Dox-Flu-SL) was successfully achieved (optimal doxorubicin:fluoxetine:lipid molar ratio of 0.02:0.05:1), obtaining a mean concentration of 257 ± 12.1 and 513 ± 29.3 µM for doxorubicin and fluoxetine, respectively. Most important, Dox-Flu-SL demonstrated drug release in synergistic ratios in cell-culture media, accounting for the improved cytotoxicity of Dox-Flu-SL over liposomal doxorubicin (LD) in both MCF-7 and MCF-7/adr cells. Pharmacokinetic studies also revealed that Dox-Flu-SL effectively prolonged drug-circulation time and reduced tissue biodistribution. Dox-Flu-SL presents a promising anticancer formulation, capable of effective reversal of drug resistance, and may constitute a novel approach for cancer therapy.

Quantification of 3-deazaneplanocin A, a novel epigenetic anticancer agent, in rat biosamples by hydrophilic interaction liquid chromatography-tandem mass spectrometric detection.

A sensitive and selective LC-MS/MS based bioanalytical method was developed and validated for the quantification of 3-deazaneplanocin A (DZNep), a novel epigenetic anti-tumor drug candidate, in Sprague-Dawley (SD) rat biosamples (plasma, urine, feces and tissue samples). The method comprises a phenylboronic acid (PBA)-containing solid phase extraction procedure, serving for binding and clean-up of DZNep in rat biosamples spiked with tubercidin (as internal standard). The analytes were separated on an Agilent hydrophilic interaction chromatography (HILIC) column. LC-MS/MS in positive ion mode was used to perform multiple reaction monitoring at m/z of 263/135 and 267/135 for DZNep and tubercidin, respectively. The limit of quantification (LOQ) of DZNep in rat biosamples was 20 ng/mL. The data of intra-day and inter-day accuracy were within 15% of nominal concentration while the precision (relative standard deviation) less than 10% for all biosamples. The extraction recoveries for DZNep and tubercidin were consistent and reproducible (around 80%) and the matrix effects were negligible (around 10% suppression) in all biosamples. This method was demonstrated to be applicable for pharmacokinetic studies of DZNep in SD rats.