This study proposes an automated approach to radiotherapy treatment planning by integrating a reinforcement-learning-style iterative framework with a multimodal Large Language Model (LLM). We specifically investigate the problem of Beam Angle Optimization, a high-dimensional and non-convex subproblem of Treatment Planning. Our system employs GPT-4V to select candidate beam angles and analyze three-dimensional dose distributions generated by Monte Carlo simulations within the MatRAD environment. Iterative plan refinement is guided by a reward function that encourages target dose conformity and penalizes excessive dose to organs at risk. We incorporate exploration-exploitation principles to strike a balance between investigating diverse action proposals and refining promising solutions. Experimental results on prostate cancer cases demonstrate that our LLM-based framework offers superior performance compared to random beam selection and can outperform the quality of deep reinforcement learning baselines, indicating the potential for LLMs to assist in complex radiotherapy treatment planning tasks.Clinical relevance-This approach is designed to alleviate the significant effort of manual treatment planning by assisting medical physicists in exploring beam configurations and systematically refining plans to improve dose coverage and protect healthy tissues.

Drug combination therapies can improve cancer treatment efficacy while overcoming drug resistance. However, with the increasing number of available drugs, predicting drug synergy within the combination space remains a challenge. For instance, current approaches often struggle with ineffective drug representations and the challenge of effectively incorporating cell line information. To address these limitations, we propose PMDSynergy, a novel drug synergy prediction framework that integrates pre-trained drug embeddings and cell line features by multi-dimensional data fusion. Leveraging the strength of pre-trained encoding techniques in terms of representational power, this method integrates drug features across three dimensions and seven omics data features, which are then integrated through a multi-modal fusion strategy based on contrastive learning to enhance the model's predictive ability. Our approach demonstrates superior performance compared to existing methods. Furthermore, we validate the robustness of our model by assessing its performance on previously unseen drug combinations, confirming its generalization capability. Overall, PMDSynergy innovatively utilizes multi-dimensional pre-trained embeddings to more accurately predict drug synergy through comprehensive representations, providing a reliable and efficient framework for drug synergy discovery. Availability: The code of our work is available at https://github.com/JieZheng-ShanghaiTech/PMDSynergy.

Despite increased surgical experience, improved perioperative care and technological advancements, postoperative ileus (POI) continuous to be a common complication following cystectomy with urinary diversion. In theory, POI can be reduced if lumen of the enteric anastomosis performed during surgery has the sufficient diameter. We aimed to evaluate different stapler types on bowel function recovery, comparing two firings of the robotic Endowrist 45 mm (EW45) stapler with the handheld EndoGia 60 mm (EG60). We hypothesized that better robotic maneuverability and a longer stapler length would create a wider anastomosis and thereby improve postoperative bowel function. This multicentre RCT was conducted between 2018 and 2021 (NCT03385798). Seventy-eight patients with bladder cancer undergoing radical cystectomy with ileal conduit were randomized 1:1 (EW45 or EG60). Time to first flatus and bowel movement was registered postoperatively and patients were followed up with the validated Bowel Function Index (BFI) preoperatively, at 3 weeks, 4 and 12 months postoperatively. The EW45 group had a higher POI rate (24% vs. 3%), longer time to first flatus (93 h vs. 71 h, p = 0.004), and longer length of hospital stay (LOS) (13 vs. 8 days, p = 0.001). No significant differences were observed in readmission, reoperation, or mortality. BFI scores did not differ between groups at any time point. Patients operated with EW45 had higher rates of POI and LOS, indicating worse outcome in the perioperative phase. However, no differences in long-term bowel function were observed and both techniques had the same risk of reoperation and mortality.

To evaluate and compare patient perceptions of artificial intelligence (AI) use in mammogram interpretation across academic and safety-net healthcare settings.

We offered a 29-item survey to patients visiting our safety-net (SNH) and academic (ACH) hospital breast imaging clinics between 04/2024-06/2024 and 02/2023-08/2023, respectively. Demographic data was compared between populations using Chi-squared tests. We used ORs (95% CI) to estimate response odds by patient factors. Significant group differences were further analyzed via multivariable regression.

A total of 924 [ACH: 518(56.1%), SNH: 406(43.9%)] surveys were collected. Participants from the ACH were older (≥ 70 years: 20%vs3.1%, p < 0.001), mostly identified as Non-Hispanic White (56%vs7.2%, p < 0.001), had higher income (≥ $100,000: 49%vs3.2%, p < 0.001), higher education (≥ college: 71%vs20%, p < 0.001) and higher self-reported knowledge of AI (68%vs56%, p < 0.001) compared to SNH. Use of AI alone or as a second reader was accepted by 74%, with SNH participants being less likely to accept [OR(95%CI): 0.71(0.53-0.96), p = 0.02]. SNH participants were more likely to request a reading by AI following radiologist-interpreted abnormalities [1.83(1.35-2.49), p < 0.001], rate AI as the same or better than a radiologist at detecting cancer [1.54(1.12-2.15), p = 0.01], and have higher concern regarding data privacy [1.87(1.22-2.93), p = 0.01]. Higher education [1.99(1.33-2.99), p < 0.001] and self-reported AI knowledge [1.98(1.38-2.83), p < 0.001] were associated with higher acceptance of AI use, while Non-Hispanic Black race [0.40(0.25-0.65), p < 0.001] was associated with lower acceptance when controlled for other covariates.

Significant differences exist in patients' views of AI between the demographically distinct academic and safety-net populations. Our study revealed lower educational attainment and Non-Hispanic Black race as independent factors associated with lower acceptance of AI.

We conducted a multi-site randomised controlled trial (RCT) of home-based exercise and self-management for patients after lung cancer surgery, which improved the intervention group's exercise capacity and self-efficacy and objectively measured physical function. We conducted this qualitative evaluation of the RCT to explore participants' experiences of the program and understand factors influencing participation, behaviour change, and acceptability.

A qualitative study nested within an RCT (ACTRN12617001283369), informed by the Theoretical Domains Framework (TDF) and Theoretical Framework of Acceptability (TFA). The intervention group participated in a 12-week postoperative remotely delivered exercise and self-management program. Following intervention maturation, consecutive intervention participants participated in individual, semi-structured telephone interviews at program discharge. Interviews were audio recorded, transcribed verbatim, cross-checked, and analysed using directed content analysis.

Twenty-one participants took part in interviews (88% of those approached). Thirty-three themes were developed from the qualitative data and mapped to TDF and TFA domains. Key TDF domains driving behaviour change included reinforcement, skills, and social influences. Examples of exercise enablers included regular physiotherapy phone consultations and activity tracking tools. Barriers to exercise included symptoms, poor weather, and competing priorities. Perceived effectiveness was a key TFA domain driving intervention acceptability. Most participants felt the program improved their physical health, exercise self-efficacy and capacity, and physical activity levels. Minor areas for improving acceptability included increased flexibility/tailoring and exercise supervision/demonstration.

The program was acceptable to participants and facilitated meaningful behaviour change. Qualitative findings aligned with quantitative RCT findings, particularly regarding improvements to exercise self-efficacy and capacity at program discharge.

Australian New Zealand Clinical Trials Registry ( http://anzctr.org.au ). Registered on 06/09/2017. Identifier: ACTRN12617001283369.

Oncologic urgent care centers (UCCs) have been shown to reduce emergency department (ED) visits. However, their impact on potentially preventable (PP) presentations, including those with diagnoses in CMS' OP-35 (OP-35) quality metric, has not been evaluated. This study assesses the impact of a specialty-specific UCC on PP ED visits in patients receiving chemotherapy.

In this retrospective, single-center analysis, patients were included if they received parenteral or oral chemotherapy within 30 days of ED presentation between March 2019 and June 2021. A UCC tailored toward managing patients with COVID-19 symptoms was opened between March 2020 and June 2021 (intervention period). Outcomes compared before and during the intervention period included weekly incidence of PP ED visits, defined as visits that would be captured in OP-35. Interrupted time series design was utilized.

A total of 2272 ED visits occurred in the study period. Most were for patients > 55 years old (n = 1706, 75%), female (n = 1227, 54%), and with gastrointestinal cancer (n = 637, 28%). Overall 928 (41%) ED visits during the study period were PP. There were 3.8 (95% CI, 0.6 to 7.0) PP weekly ED visits during the intervention period versus 11.0 (95% CI, 9.4 to 12.6) before the intervention period, reflecting a reduction in 7.2 weekly visits (95% CI, - 10.8 to - 3.6).

The opening of an oncologic UCC was associated with a reduction in PP ED visits. UCCs can be important in an oncologic service line to reduce unnecessary ED usage, but prospective evaluations are needed to confirm this finding.

Chemoradiotherapy (CRT) for head and neck cancer (HNC) often leads to dysphagia and myelosuppression (e.g., neutropenia and anemia). This study investigated the associations between myelosuppression, muscle mass changes, and dysphagia in patients undergoing CRT for HNC.

This retrospective observational study included 136 patients with HNC who underwent curative CRT between January 2019 and August 2024. Anemia (hemoglobin ≤ 8.0 g/dL) and neutropenia (absolute neutrophil count < 1000/mm³) were classified based on routine blood tests during CRT. The primary outcome was feeding tube dependency 3 months after completing CRT. Swallowing muscle mass was evaluated using the cross-sectional area of the tongue complex muscles (TCM).

On multivariate analysis, anemia was significantly associated with a decline in the TCM cross-sectional area (coefficient: 8.21, P < 0.001) and feeding tube dependency (odds ratio: 5.25, P = 0.02). In contrast, neutropenia had no significant impact on either.

Anemia, an adverse effect of CRT for HNC, was an independent risk factor for feeding tube dependency at 3 months post-CRT and was significantly associated with a decline in the TCM cross-sectional area.

Gemcitabine-cisplatin doublet is a standard first-line regimen for metastatic gallbladder cancer (GBC), though prospective real-world data remains scarce. We evaluated the efficacy, safety, and prognostic factors in the North Indian patients.

Between March 2021 and December 2022, all patients with histologically proven metastatic GBC were prospectively enrolled. Eligible patients had ECOG 1-2 and adequate organ function. Gemcitabine 1000 mg/m² (days 1, 8) and cisplatin 75 mg/m² (days 1-2) were given every 3 weeks for up to 6 cycles, until progression or intolerance. Patients receiving ≥ 2 cycles were evaluable for efficacy. Responses were assessed by RECIST 1.1. Kaplan-Meier, univariate, and multivariate analyses were performed.

Sixty-three patients were included in the study (Mean Age 56 years; 62% female). ORR was 63.5%; DCR 84.5%. Median PFS was 5.0 months; median OS 11.0 months. Six-month OS was 82.1%, 12-month OS 36.9%. Independent predictors of PFS included % change in CA19-9, platelet-lymphocyte ratio, alkaline phosphatase, and extra-abdominal disease. Neutropenia (any grade 73%, grade 3-4: 27%) was the most common toxicity; G-CSF was used therapeutically in 41%. Non-hematological AEs were mild-to-moderate. No treatment-related deaths occurred.

Gemcitabine-cisplatin remains effective and tolerable in metastatic GBC. Despite the emergence of Gemcitabine-Cisplatin with durvalumab as new standard, this prospective dataset provides valuable real-world outcomes from a high-incidence region with limited access to immunotherapy.

Gynecologists are the main group promoting vaccines. Previous studies have paid less attention to their human papillomavirus(HPV) vaccination status.Understanding their acceptance and true attitudes toward vaccines, as well as the reasons for not getting vaccinated, is of great significance for promoting vaccines. A cross-sectional survey was conducted from June to August 2024 across six Chinese cities by the Cancer Endocrinology Committee of the Chinese Anti-Cancer Association, using a convenience sampling strategy to distribute questionnaires to gynecologists in tertiary public hospitals. Results were presented as case numbers (percentages), with group comparisons made using either the χ² test or Fisher's exact probability method; influencing factors were analyzed via logistic regression. Out of 577 collected questionnaires, 563 were valid. The vaccination rate among female gynecologists was 46.07% (223/484), with the quadrivalent HPV vaccine being most common at 40.36%. The primary reason for not vaccinating was exceeding age limits, followed by male gender. And 12.35% of gynecologists were reluctant to vaccinate due to annual cervical cancer screening. 83.13% (468/563) gynecologists supported male vaccination.The level of support for vaccinating underage daughters was significantly higher than that for sons (87.74% vs. 68.92%, p<.001). It is essential to enhance public awareness of the protective benefits of HPV vaccination for males and its efficacy in preventing various diseases beyond cervical cancer.

Despite advancements in healthcare data management, missing data in electronic health records (EHR) and patient-reported outcomes remain a persistent challenge, limiting their usability in healthcare analytics. Conventional imputation methods often struggle to capture complex nonlinear relationships, require extensive computation time, and are limited in addressing various types of missing data mechanisms. To overcome these challenges, we propose the clustering-informed shared-structure variational autoencoder (CISS-VAE), which utilizes the strengths of Bayesian neural networks. This model can effectively capture complex associations and accommodate various missing data mechanisms, including missing not at random (MNAR). We also develop iterative learning algorithms that further enhance missing data imputation accuracy while preventing overfitting. Comprehensive simulations demonstrate the superior accuracy of our model compared to traditional and contemporary methods. We apply our method to EHR data from early-stage breast cancer patients at Memorial Sloan Kettering Cancer Center, aiming to mitigate the impact of missing data and enhance health monitoring and analyses.