Health system-scale language models are all-purpose prediction engines.

Physicians make critical time-constrained decisions every day. Clinical predictive models can help physicians and administrators make decisions by forecasting clinical and operational events. Existing structured data-based clinical predictive models have limited use in everyday practice owing to complexity in data processing, as well as model development and deployment1-3. Here we show that unstructured clinical notes from the electronic health record can enable the training of clinical language models, which can be used as all-purpose clinical predictive engines with low-resistance development and deployment. Our approach leverages recent advances in natural language processing4,5 to train a large language model for medical language (NYUTron) and subsequently fine-tune it across a wide range of clinical and operational predictive tasks. We evaluated our approach within our health system for five such tasks: 30-day all-cause readmission prediction, in-hospital mortality prediction, comorbidity index prediction, length of stay prediction, and insurance denial prediction. We show that NYUTron has an area under the curve (AUC) of 78.7-94.9%, with an improvement of 5.36-14.7% in the AUC compared with traditional models. We additionally demonstrate the benefits of pretraining with clinical text, the potential for increasing generalizability to different sites through fine-tuning and the full deployment of our system in a prospective, single-arm trial. These results show the potential for using clinical language models in medicine to read alongside physicians and provide guidance at the point of care.

Brusatol improves the efficacy of sorafenib in Huh7 cells via ferroptosis resistance dependent Nrf2 signaling pathway.

BACKGROUND: Hepatocellular carcinoma (HCC) is a common malignancy with a poor prognosis. The recommended treatment of unresectable HCC involves targeted therapy, for example sorafenib, combined with immunotherapy. A recent article reported that sorafenib could induce ferroptosis escape in HCC. Brusatol is a novel Nrf2 inhibitor that takes effects in various diseases. In our study, we aimed to identify whether the addition of Brusatol to sorafenib could reverse ferroptosis escape in Huh7 cells. METHODS: The cultured Huh7 cells treated by sorafenib with or without Brusatol addition were harvested for ferroptotic phenotype experiments and ferroptosis-related markers such as GPX4 and SLC7A11 were detected. In vivo experiments were conducted to discover the effect of Brusatol in combination with sorafenib in liver tumor bearing mice. Mechanism signaling pathways were detected by RNA-sequencing. RESULTS: Brusatol alone could induce Huh7 cell death and sorafenib could moderately mediate Huh7 cell ferroptosis by paradoxically inhibiting GPX4. However, sorafenib simultaneously upregulates Nrf2 signaling in Huh7 cells fighting against ferroptosis to result in sorafenib resistance. The addition of Brusatol could potentiate ferroptosis in Huh7 cells through downregulating Nrf2 and the downstream HO-1 and NQO1, thus enhancing the efficacy of sorafenib, which could be reversed by ferrostatin-1 treatment. CONCLUSION: In conclusion, Brusatol improves the efficacy of sorafenib by inducing ferroptosis via hindering Nrf2 signaling activation in HCC.

Treatment of Superficial Incision Dehiscence after Abdominal Surgery by Z-Plasty: A Retrospective Case Series.

OBJECTIVE: To explore the treatment of superficial incision dehiscence after abdominal surgery by Z-plasty combined with negative-pressure wound therapy. METHODS: A retrospective study was performed on seven patients with superficial abdominal incision dehiscence from October 2018 to February 2019. All patients were given systemic antibiotics and nutrition support. During the first stage, surgical debridement with negative-pressure wound therapy was performed. Local Z-plasty was performed in the second stage. RESULTS: The incision healed well in all patients, and no infection or necrosis occurred in the flaps. During the follow-up of 7.3 months (range, 5-10 months), no incision rupture or redehiscence occurred. CONCLUSIONS: Surgical debridement, negative-pressure wound therapy, and Z-plasty can be used to treat superficial abdominal incision dehiscence and achieve good therapeutic effect and prognosis. Z-plasty can be used as an alternative to direct suture of incisions because of its simplicity and excellent results.

Staged Treatment of Chest Wall Radiation-Induced Ulcer With Negative Pressure Wound Therapy and Latissimus Dorsi Myocutaneous Flap Transplantation.

Chest wall ulcer induced by postmastectomy radiation therapy (PRMT) remains challenging for plastic surgeons because of the reduced blood supply, fibrosis, and impaired cellular potential in the irradiated area. In this study, chest wall ulcer was treated with negative pressure wound therapy (NPWT) and the latissimus dorsi myocutaneous (LDM) flap reconstruction in 2 stages. A retrospective study was performed on consecutive patients with chronic radiation-induced ulcers in chest wall from June 2012 to June 2017. Surgical debridement and NPWT were performed in the first stage and the chest wall reconstructed by the LDM flap transplantation after extensive debridement in the second stage. There were 10 female patients with chest wall ulcers with a mean age of 60.3 years. The average duration of the ulcers was 21.2 months and the ulcers varied from 1 × 2 to 5 × 7 cm. Histological examination denied any recurrent breast cancer or radiation-related malignancy. Negative pressure wound therapy was applied with 100 to 125 mm Hg negative pressure during a period of 5 to 7 days in the first stage. The LDM flap varied from 11 × 15 to 15 × 20 cm. The mean follow-up was 25.9 months. All the flaps survived well with satisfactory appearance and there was no donor-site morbidity or ulcer recurrence during the follow-up period. The staged treatment of the chest wall radiation ulcer incorporated the benefits of NPWT and LDM flap. It is beneficial in increasing the blood and nutrient supply to the irradiated tissue, enhancing the debridement and promoting tissue healing, thus improving the flap survival and decreasing the ulcer recurrence.

MCM4 potentiates evasion of hepatocellular carcinoma from sorafenib-induced ferroptosis through Nrf2 signaling pathway.

Hepatocellular carcinoma (HCC) is one of the most common malignancies worldwide. It poses an enormous socioeconomic burden and is a serious public health threat globally due to its poor prognosis. Ferroptosis is a newly identified non-apoptotic form of cell death characterized by lipid peroxidation, iron accumulation, and reactive oxygen species (ROS) generation. However, tumor cells have evolved diverse mechanisms to evade ferroptosis, conferring resistance to drugs. Sorafenib, a first-line therapy for advanced HCC, triggers ferroptosis by selectively targeting solute carrier family 7 member 11 (SLC7A11) to deplete glutathione and inhibit glutathione peroxidase 4 (GPX4), thereby effectively eliminating tumor cells. However, sorafenib resistance has been widely reported, and the precise mechanisms underlying sorafenib drug resistance remain unclear. The minichromosome maintenance (MCM) protein family contains 10 members with vital roles in DNA replication and cell cycle progression. MCM4, a member of the MCM protein family, might be a potential biomarker in pan-cancer analysis. The present study found that MCM4 was upregulated in liver cancer using bioinformatics analysis and sorafenib-treated HCC cells. Moreover, MCM4 might be regarded as a prognostic biomarker for HCC. Further experiments revealed that MCM4-inhibition enhanced the efficacy of sorafenib through elevation of ferroptosis both in vitro and in vivo. Mechanistically, MCM4 potentiates sorafenib-induced ferroptosis evasion in HCC by promoting nuclear factor erythroid 2-related factor 2 (Nrf2) signaling activation. However, no direct interactions were found between Nrf2 and MCM4. Overall, these findings suggest a potential therapeutic strategy for HCC by targeting MCM4 inhibition.

Assessing superficial temporal artery-middle cerebral artery anastomosis patency using FLOW 800 hemodynamics.

OBJECTIVE: The objective of this study was to investigate the use of indocyanine green videoangiography with FLOW 800 hemodynamic parameters intraoperatively during superficial temporal artery-middle cerebral artery (STA-MCA) bypass surgery to predict patency prior to anastomosis performance. METHODS: A retrospective and exploratory data analysis was conducted using FLOW 800 software prior to anastomosis to assess four regions of interest (ROIs; proximal and distal recipients and adjacent and remote gyri) for four hemodynamic parameters (speed, delay, rise time, and time to peak). Medical records were used to classify patients into flow and no-flow groups based on immediate or perioperative anastomosis patency. Hemodynamic parameters were compared using univariate and multivariate analyses. Principal component analysis was used to identify high risk of no flow (HRnf) and low risk of no flow (LRnf) groups, correlated with prospective angiographic follow-ups. Machine learning models were fitted to predict patency using FLOW 800 features, and the a posteriori effect of complication risk of those features was computed. RESULTS: A total of 39 cases underwent STA-MCA bypass surgery with complete FLOW 800 data collection. Thirty-five cases demonstrated flow after anastomosis revascularization and were compared with 4 cases with no flow after revascularization. Proximal and distal recipient speeds were significantly different between the no-flow and flow groups (proximal: 238.3 ± 120.8 and 138.5 ± 93.6, respectively [p < 0.001]; distal: 241.0 ± 117.0 and 142.1 ± 103.8, respectively [p < 0.05]). Based on principal component analysis, the HRnf group (n = 10) was characterized by high-flow speed (> 75th percentile) in all ROIs, whereas the LRnf group (n = 10) had contrasting patterns. In prospective long-term follow-up, 6 of 9 cases in the HRnf group, including the original no-flow cases, had no or low flow, whereas 8 of 8 cases in the LRnf group maintained robust flow. Machine learning models predicted patency failure with a mean F1 score of 0.930 and consistently relied on proximal recipient speed as the most important feature. Computation of posterior likelihood showed a 95.29% chance of patients having long-term patency given a lower proximal speed. CONCLUSIONS: These results suggest that a high proximal speed measured in the recipient vessel prior to anastomosis can elevate the risk of perioperative no flow and long-term reduction of flow. With an increased dataset size, continued FLOW 800-based ROI metric analysis could be used to guide intraoperative anastomosis site selection prior to anastomosis and predict patency outcome.

A Spatial-Temporal Graph Attention Network for Automated Detection and Width Estimation of Cortical Spreading Depression Using Scalp EEG.

We present an end-to-end Spatial-Temporal Graph Attention Network (STGAT) for non-invasive detection and width estimation of Cortical Spreading Depressions (CSDs) on scalp electroencephalography (EEG). Our algorithm, that we refer to as CSD Spatial-temporal graph attention network or CSD-STGAT, is trained and tested on simulated CSDs with varying width and speed ranges. Using high-density EEG, CSD-STGAT achieves less than 10.96% normalized width estimation error for narrow CSDs, with an average normalized error of 6.35%±3.08% across all widths, enabling non-invasive and automated estimation of the width of CSDs for the first time. In addition, CSD-STGAT learns the temporal and spatial features of CSDs simultaneously, which improves the "spatio-temporal tracking accuracy" (i.e., the defined detection performance metric at each electrode) of the narrow CSDs by up to 14%, compared to the state-of-the-art CSD-SpArC algorithm, with only one-tenth of the network size. CSD-STGAT achieves the best spatio-temporal tracking accuracy of 86.27%±0.53% for wide CSDs using high-density EEG, which is comparable to the performance of CSD-SpArC with less than 0.38% performance reduction. We further stitch the detections across all electrodes and over time to evaluate the "temporal accuracy". Our algorithm achieves less than 0.7% false positive rate in the simulated dataset with inter-CSD intervals ranging from 5 to 60 minutes. The lightweight architecture of CSD-STGAT paves the way towards real-time detection and parameter estimation of these waves in the brain, with significant clinical impact.