JCO Clinical Cancer Informatics最新文献

Purpose: Direct pediatric information to inform chemotherapy emetogenicity in pediatric patients is limited. Therefore, the framework for antiemetic selection is uncertain. This study classified the acute emetogenicity of chemotherapy regimens in pediatric patients using data extracted from the electronic health record (EHR).

Methods: This retrospective, single-institution study extracted data from the EHR of patients age 0 to 18 years who received chemotherapy during an inpatient admission from July 1, 2018, through February 29, 2024. Data were organized by patient and chemotherapy block including patient demographics; date, time, and route of chemotherapy and antiemetic administration; and date and time of vomiting. When at least 30 patients received the same chemotherapy and antiemetics during a chemotherapy block, the proportion of chemotherapy blocks where patients experienced complete, partial, or failed chemotherapy-induced vomiting control was determined. Chemotherapy regimen emetogenicity was assigned using a revision of an accepted pediatric chemotherapy emetogenicity classification framework that adjusted for antiemetic administration.

Results: Seven thousand two hundred ninety-six chemotherapy blocks in 1,386 patients were identified. The emetogenicity of 25 chemotherapy regimens was classified: highly (7), moderately (5), low (10), and minimally (3) emetogenic. For 19 of these, no direct pediatric information was previously available. In five, our findings confirm the previous pediatric emetogenicity classification. Relative to emetogenicity classifications for adults, our findings led to classifications that were higher (seven regimens), lower (one regimen), or the same (four regimens).

Conclusion: We have applied a novel method, EHR data extraction, to provide direct pediatric evidence to classify chemotherapy emetogenicity. Increasing the certainty of chemotherapy emetogenicity facilitates effective antiemetic selection for pediatric patients. This method may be applied in multi-institution studies to increase the number of chemotherapy regimens whose emetogenicity is classified using direct pediatric evidence.

Purpose: Cyclin-dependent kinase 4 and 6 inhibitors (CDKIs) are effective breast cancer therapies but pose adherence challenges because of cost, side effects, and complexity of medication schedule. We assessed the feasibility and usability of a smart label-enabled remote therapeutic monitoring (RTM) mHealth intervention for women with breast cancer prescribed a CDKI. Exploratory adjusted analyses examined factors associated with usability and CDKI adherence.

Methods: Participants were recruited from a comprehensive cancer center between April and August 2024. For 3 months, participants used Tappt smart labels and web app to record CDKI doses, receive missed dose reminders, report symptoms biweekly, and complete baseline and follow-up surveys. Alerts were sent to oncology teams for nonadherence (>20% missed doses) or moderate-to-severe symptoms. Feasibility was defined as ≥70% of participants using the smart label >30 days and completing the follow-up survey. Usability was assessed using the System Usability Scale, with a benchmark score of ≥68. Linear regression was used to examine factors associated with usability and CDKI adherence.

Results: Among 168 screened, 107 were eligible and reached; 75.7% (81/107) consented; 90.1% (73/81) completed the follow-up survey, and 88.9% (72/81) used the intervention >30 days. Most participants self-identified as White (69.9%), were privately insured (72.6%), and had early-stage breast cancer (58.9%) and depression or anxiety (58.9%). The mean usability score was 75.8; participants who self-identified as Black reported 12.0 points higher usability than those who self-identified as White (P = .03). Mean CDKI adherence was 92.8%. A history of anxiety or depression was associated with an 8.6 percentage-point lower CDKI adherence rate (P = .02).

Conclusion: A smart label-enabled RTM mHealth intervention exceeded feasibility and usability benchmarks and showed promise for supporting CDKI adherence and symptom management.

Purpose: Chain-of-thought prompting is a method to make large language models generate intermediate reasoning steps when solving a complex problem. OpenAI's o1 preview and GPT-5 have been trained to create such a chain of thought internally before giving a response and have been claimed to surpass various benchmarks requiring complex reasoning. The purpose of this study was to evaluate their performance in text mining in oncology.

Methods: Six hundred trials from high-impact medical journals were classified depending on whether they allowed for the inclusion of patients with localized and/or metastatic disease. GPT-4o, o1 preview, and GPT-5 at different reasoning effort settings were instructed to do the same classification based on the publications' abstracts.

Results: For predicting whether patients with localized disease were enrolled, GPT-4o and o1 preview achieved F1 scores of 0.80 (0.76-0.83) and 0.91 (0.89-0.94), respectively. For predicting whether patients with metastatic disease were enrolled, GPT-4o and o1 preview achieved F1 scores of 0.97 (0.95-0.98) and 0.99 (0.99-1.00), respectively. For GPT-5, the F1 scores for predicting the eligibility of patients with localized disease increased from 0.84 to 0.93 and 0.94 with increased reasoning effort. F1 scores for metastatic disease were 0.97, 0.99, and 0.99.

Conclusion: o1 preview outperformed GPT-4o in extracting if people with localized and/or metastatic disease were eligible for a trial from its abstract. GPT-5 at high reasoning effort settings outperformed both GPT-4o and o1 preview, supporting the notion that reasoning models could become the new standard for text mining in medicine.

Purpose: Multidisciplinary breast tumor boards (MTBs) are essential for optimizing breast cancer treatment but face challenges related to logistics, variability in expertise, and lack of standardization. Large language models may support clinical decision making. This study evaluates the accuracy of adjuvant therapy recommendations generated by an artificial intelligence (AI)-driven tool, TheSerenityBot (TSB), in comparison with Claude-2 and GPT-4, using expert MTB consensus as the reference.

Methods: Postoperative breast cancer cases reviewed at the National University Hospital, Singapore, between June and November 2023 were retrospectively analyzed. Eligible patients were women with invasive or preinvasive breast cancer who underwent surgery. Metastatic cases were excluded. TSB, a Claude-2-based model augmented with 2023 National Comprehensive Cancer Network guidelines, generated adjuvant therapy recommendations across seven treatment modalities. Outputs from TSB, Claude-2, and GPT-4 were evaluated for concordance with MTB recommendations. Model performance was assessed using generalized estimating equations.

Results: Fifty patients were included (mean age, 59.8 years); 75.5% had hormone receptor-positive tumors, and 60.0% underwent breast-conserving surgery. TSB demonstrated the highest overall accuracy (0.89), followed by Claude-2 (0.86) and GPT-4 (0.78). GPT-4 showed significantly lower accuracy in genetic testing recommendations (odds ratio [OR], 0.05 [95% CI, 0.015 to 0.149]; P < .001), whereas Claude-2 was less accurate in radiotherapy recommendations (OR, 0.41 [95% CI, 0.17 to 0.98]; P = .040).

Conclusion: A guideline-augmented AI tool such as TSB shows promise in supporting adjuvant therapy decisions in breast cancer. To improve clinical relevance, future iterations will incorporate individualized patient factors, broader guideline frameworks, and electronic health record integration. Prospective trials are ongoing to assess the real-world impact.

Purpose: Results from single-arm clinical trials can be contextualized by comparing against external controls (ECs) derived from real-world data (RWD). However, lack of randomization and differences in variable capture between data sources may introduce bias into estimates of treatment effect and standard error, the extent of which can be assessed via meta-analysis of comparisons between clinical trial control arms and their EC replicates.

Methods: Clinical trial progression-free survival (PFS) outcomes from the 14 chemotherapy control arms of 12 non-small cell lung cancer clinical trials were replicated using the US nationwide deidentified Flatiron Health electronic health record-derived database, with real-world PFS (rwPFS) as the end point. A meta-analysis of log_e hazard ratios (HRs) comparing randomized controlled trial (RCT) and RWD control arms was conducted. For illustration, the meta-analysis results were used to restore correct operating characteristics of a hypothetical prospective single-arm study with EC.

Results: With the exception of one outlier, rwPFS outcomes were on average similar to PFS outcomes, albeit with substantial between-study variation. RCT compared with RWD arms differed by a mean log_e HR of -0.001, with a standard deviation of 0.164 (including the outlier). Applying these estimates to adjust error probabilities in a hypothetical prospective EC study revealed that between-study variation of bias in this setting should be adjusted for, to avoid incorrect decision making.

Conclusion: The close alignment of results between RCT and RWD increases confidence that RWD ECs using the rwPFS end point in this disease setting can provide context for future single-arm clinical trials despite potential differences in end point assessment.

Purpose: Lung cancer is the leading global cause of cancer mortality with substantial evidence of inequity, disparity in process and outcomes, and unwarranted clinical variation. Over the last decades, there has been major evolution and discovery in best evidence-based practice (EBP), enhancing diagnostics, management, and the delivery of precision medicine. However, questions remain about the completeness of translation of best EBP into delivered care.

Design: Learning health systems (LHSs) have been defined as improvement environments where knowledge generation processes are embedded into daily clinical practice to continually improve the quality, safety, and outcomes of health care delivery. Lung cancer clinical quality registries (CQRs) provide a rigorous infrastructure supporting LHS function through the collection, analysis, and reporting of care process and outcome information delivered by health service organizations. CQRs measure the appropriateness and effectiveness of delivered care and report on the degree of best EBP delivery by stakeholder providers. The provision of risk-adjusted, benchmark reporting to stakeholders describes equity, disparity, and unwarranted clinical variation and is a fundamental driver of improvement in the safety and quality of care provided to consumers.

Results: There is mounting international evidence of the positive impacts of CQR reporting on management processes, health care infrastructure, survival, quality improvement, and education within lung cancer communities. The use of implementation science approaches including the Knowledge to Action framework targets bridging the gaps between evidence-based knowledge and practice.

Conclusion: Registry evolution is exampled by the Danish Lung Cancer Registry, National Lung Cancer Audit (United Kingdom), Dutch Lung Cancer Audit, and Victorian Lung Cancer Registry (Australia), which identify innovation opportunities to close the evidence-practice gap, overcome service deficits, and lead to better decision making for health care improvement.

Purpose: Real-world data (RWD) are critical for breast cancer (BC) research but are limited by privacy concerns, missing information, and data fragmentation. This study explores synthetic data (SD) generated through advanced generative models to address these challenges and create harmonized longitudinal data sets.

Methods: A data set of 1052 patients with human epidermal growth factor receptor 2-positive and triple-negative BC from the Informatics for Integrating Biology and the Bedside (i2b2) platform was used. Advanced generative models, including generative adversarial networks (GANs), variational autoencoders (VAEs), and language models (LMs), were applied to generate synthetic longitudinal data sets replicating disease progression, treatment patterns, and clinical outcomes. The Synthethic Validation Framework (SAFE) powered by Train was used to evaluate the fidelity, utility, and privacy. SD were tested across three settings: (1) integration with i2b2 for privacy-preserving data sets; (2) multistate disease modeling to predict clinical outcomes; and (3) generation of synthetic control groups for clinical trials.

Results: The synthetic data sets exhibited high fidelity (score 0.94) and ensured privacy, with temporal patterns validated through time-series analyses and Uniform Manifold Approximation and Projection embeddings. In setting A, SD accurately mirrored RWD on the i2b2 platform while maintaining privacy. In setting B, incorporating SD improved the predictive performance of a multistate disease progression model, increasing the C-index by up to 10%. In setting C, SD replicated the end points of the APT trial, demonstrating its feasibility for generating synthetic control arms with preserved statistical properties of the real data set.

Conclusion: AI-generated longitudinal SD effectively address key challenges in RWD use in BC. This approach can improve translational research and clinical trial design while ensuring robust privacy protection. Integration with platforms such as i2b2 highlights their scalability and potential for broader applications in oncology.

Purpose: We developed and validated a dual-purpose, open-access Rural-Urban Commuting Area (RUCA) tool to standardize geographic coding for cancer disparities research, addressing National Institutes of Health (NIH) Helping to End Addiction Long-term (HEAL) Initiative Common Data Element requirements while supporting institutional catchment area analyses.

Methods: This web-based tool¹⁶ integrates US Department of Agriculture RUCA codes with census tract data and electronic health record systems, meeting NIH HEAL Initiative Findable, Accessible, Interoperable, and Reusable (FAIR) data ecosystem requirements. We implemented the tool using Wake Forest Cancer Center's 2023 registry data (n = 21,219) and conducted systematic comparison with county-level Rural-Urban Continuum Code (RUCC) classifications using 18,714 cancer cases across 336 ZIP codes, focusing on breast, colon, and lung cancers to demonstrate enhanced geographic granularity.

Results: Among 21,219 patients with cancer, 19.51% (n = 4,140) resided in rural areas, with 4.81% (n = 1,022) in the most rural census tracts (RUCA codes 7-10). Comparative analysis revealed 9.4% disagreement between RUCA and RUCC classifications, affecting 1,765 patients. Twenty-eight ZIP codes classified as rural by RUCA were located within metropolitan counties according to RUCC, encompassing 109 patients with cancer who would be misclassified using county-level measures. As a separate use case, integration with NIH HEAL Initiative standardized rurality data collection across 15 research studies.

Conclusion: The RUCA tool addresses critical gaps in geographic data standardization by providing census tract-level precision that county-level classifications miss. This dual-application framework aligns institutional catchment analyses with national standardization efforts, identifying 109 patients with cancer who would be misclassified as urban residents using traditional county-level approaches, thereby enhancing targeted interventions for rural cancer care access.

Purpose: Patients with advanced cancer often experience pain symptoms. Pain-cognitive behavioral therapy (pain-CBT) represents an effective psychological treatment for chronic pain, yet access remains limited. We conducted a pilot study to assess the feasibility and acceptability of a mobile health (mHealth) intervention that integrates pain-CBT with opioid education and tracking to improve chronic pain management in patients with advanced cancer.

Methods: Adults with advanced cancer and pain (≥4/10, Numeric Rating Scale) using opioids tested the smartphone-based intervention for 28 days, completed baseline, end-of-study, and 2-week postintervention surveys, and participated in optional qualitative interviews. The intervention assessed pain, mood, catastrophizing, sleep, and opioid use, and provided tailored just-in-time adaptive interventions, and daily psychoeducation (articles, serious game). We assessed feasibility (≥50% app-use), acceptability (acceptability E-scale), and pre-post intervention changes in pain, and conducted thematic analysis of perceived impact and usefulness.

Results: Among 64 eligible patients, 32 (mean age, 55.41 years; 55% female; 32% rural-dwelling) enrolled. Of those, 59% (n = 19) used the app ≥50% of days on study, and rated the intervention with good acceptability (mean, 24.85; standard deviation, 3.72). Nonsignificant reductions in pain intensity, pain interference, and pain catastrophizing were observed from baseline to 4- and 6-week follow-ups. In debriefing interviews, patients described that the intervention contributed to pain self-management knowledge, promoted pain coping skills, and reduced opioid stigma.

Conclusion: Study results support feasibility and acceptability of a pain-CBT intervention for patients with advanced cancer pain. Although exploratory analyses showed nonsignificant improvements in pain outcomes, qualitative findings indicate meaningful engagement and skill development. Future testing is needed to determine intervention efficacy.