Background: Estimating operative time with low accuracy and precision leads to suboptimal scheduling of available operating rooms (ORs) and waste of healthcare resources. We aimed to develop a predictive machine learning (ML) model for operative time using our institutional data to improve the scheduling of neurosurgical ORs, minimize waste, and maximize resource utilization. Methods: We developed a predictive model using our institution’s multicenter registry, which included diverse neurosurgical cases across three tertiary hospitals. We applied multiple ML techniques, including linear regression, support vector regression, deep neural networks, and extreme gradient boosting regression (XGBR). The best model was combined with the first fit decreasing bin-packing algorithm to create a practical OR scheduling system based on prior data. Results: The XGBR model exhibited the best performance, with lower mean absolute errors (MAEs) (root mean squared error: 52.24, MAE: 37.25) and higher R2 (0.76) values than other models. Implementing this model in a simulated OR scheduling scenario using institutional data, we successfully scheduled 30 random procedures with minimal residual time, such that the model ran out of procedures for the week with 86 and 171 min remaining in the cranial and spinal ORs, respectively, demonstrating potential efficiency gains. The cranial procedures were scheduled over 4 days and spinal procedures over 6 days, optimizing available OR time. Conclusion: Integrating ML models into operative time prediction can significantly improve the accuracy and efficiency of surgical schedules. This approach minimizes wasted OR time and enhances resource allocation, potentially improving patient outcomes and satisfaction by reducing waiting times.
Role of a Deep-Learning based Convolutional Neural Network Model for Real-Time Ventricular Tachycardia Alarm Classification
Ventricular tachycardia (VT) is a life-threatening arrhythmia that requires both relatively rapid and accurate detection in intensive care units (ICUs). Continuous monitoring systems play a crucial role in detecting them. However, previous studies have reported that nearly 9 out of 10 arrhythmia alarms in ICUs tend to be false positives, which usually transpire to a well-documented phenomenon called “alarm fatigue” that leads to desensitization, delayed responses and increased cognitive burden on healthcare providers. We developed a deep learning based, one-dimensional convolutional neural network (1D-CNN) to classify VT alarms using multiple raw waveform inputs, including two electrocardiogram (ECG) leads, photoplethysmogram (PPG) and arterial blood pressure (ABP) signals. The model was trained using the publicly available VTaC Arrhythmia Benchmark Dataset. Ten-second waveform segments preceding each alarm were pre-processed and used to train the model to correctly classify the VT alarm. On the test set, the model achieved an area under the receiver operating characteristic curve (AUC-ROC) of 0.901, overall accuracy of 83.22%, F1-score of 73.3%, sensitivity of 77.53%, specificity of 85.63%, and positive predictive value (PPV) of 69.57%. The model successfully detected over three-quarters of them while significantly reducing false positive rates for the detection of VT. This study demonstrates that a deep learning based 1D-CNN model using short segments of raw waveform data can achieve robust performance in distinguishing true and false VT alarms.
Standardizing Antimicrobial Prophylaxis in Pediatric Patients Requiring Extracorporeal Membrane Oxygenation
Background: Prophylactic antimicrobials for pediatric patients requiring extracorporeal membrane oxygenation (ECMO) support are not well delineated; there are no consistent guidelines nationally or internationally. We analyzed antimicrobial use before and after implementing a narrow-spectra ECMO antimicrobial prophylaxis protocol intended to standardize prophylaxis by risk related to ECMO type and configuration. Methods: This retrospective, single-center study included children under 18 years old requiring ECMO between May 2018 and December 2021. Medical and cardiac surgical patients requiring open-chested (central) or peripheral veno-venous or veno-arterial ECMO were included. We evaluated the impact of an ECMO antimicrobial prophylaxis protocol by comparing the appropriateness of antimicrobial use as well as nosocomial infections in the pre versus post protocol. Results: Our study included 55 children with 58 individual ECMO episodes. More than 50% of the patients were open-chested during ECMO support, and the majority of ECMO support was veno-arterial. In the post protocol period, patients were 4 times as likely to have appropriate antimicrobial prescribing (Odds ratio 4.4; 95% confidence interval: 1.1–19; P = 0.036), including both prophylactic (per protocol) and therapeutic (chart documentation of an infection with appropriate therapeutic antimicrobials, including drug and duration). There was no difference in the odds of nosocomial infection occurring during ECMO pre protocol versus post protocol (pre vs. post protocol, P = 0.71 for invasive infections, P = 0.73 for all infections including potential colonization). Conclusions: After implementation of a standardized narrow-spectra ECMO antimicrobial prophylaxis protocol, we observed an increase in appropriate and standardized antimicrobial prescribing without concomitant increase in nosocomial infections.
Cognitive Function in People With Pediatric Multiple Sclerosis Over 2 Years
Primary Results From Blood and Marrow Transplant Clinical Trials Network 1702: Clinical Transplant-Related Long-Term Outcomes of Alternative Donor Allogeneic Transplantation
PURPOSE The likelihood of finding a human leukocyte antigen (HLA)–matched unrelated donor (MUD) for hematopoietic cell transplantation can be predicted using a donor search prognosis score. Patients without a MUD may use alternative donors (haploidentical related, mismatched unrelated, or umbilical cord blood). METHODS This multicenter biological assignment trial was conducted by the Blood and Marrow Transplant Clinical Trials Network (BMT CTN 1702). Eligibility criteria were broad to mirror clinical practice. The primary end point was 2-year survival from evaluability and compared between those Very Likely (>90%) and Very Unlikely (<10%) to find a MUD. All other patients, Less Likely to find a MUD, were enrolled in an observational arm. Transplant outcomes were compared for all three groups. RESULTS A total of 1,751 evaluable patients at 47 centers were Very Likely (54.7%), Less Likely (29.5%), and Very Unlikely (15.8%) to identify a MUD. Survival did not differ in univariate (hazard ratio [HR], 1.00 [95% CI, 0.82 to 1.21]; P = .98) or multivariate (HR, 1.07 [95% CI, 0.86 to 1.33]; P = .56) analyses between the Very Unlikely and Very Likely groups, measured through 2 years from the beginning of a search for a MUD or alternative donor. Of the transplanted patients (n = 1,179), MUD was used for 94% of the Very Likely, 38% of Less Likely, and 9% of Very Unlikely patients. Multivariate analyses showed no differences in relapse, treatment-related mortality, disease-free survival, and acute and chronic graft-versus-host diseases for the three search prognosis groups after transplantation. CONCLUSION Using a donor search prognosis strategy to prioritize an alternative donor for patients Very Unlikely to find a MUD resulted in survival and transplant outcomes that were not statistically different compared with those Very Likely to find a MUD.
Development and validation of a gene expression signature to predict early events in patients with follicular lymphoma
Abstract Although follicular lymphoma (FL) typically follows an indolent course, patients with FL who experience early events, such as transformation or progression, have increased risk of death related to lymphoma. The FL24Cx is an algorithm based on a 45-target gene expression profiling (GEP) assay, which was developed and trained using 265 formalin-fixed, paraffin-embedded tissue samples on a reliable platform to predict, at the time of diagnosis, whether a patient will experience an event within 24 months. The modeling also confirmed and relied upon previously reported synergy between immune response (IR) gene expression signatures IR1 and IR2. Once locked, the 5-factor logistic regression FL24Cx model was independently validated in a retrospectively assessed cohort of 232 patients from 2 immunochemotherapy-treated arms of SWOG Cancer Research Network S0016 phase 3 clinical trial, in which it assigned 169 patients to the low-risk group with 29 events before 24 months (17.2%) and 63 patients to the high-risk group with 24 events before 24 months (38.1%). The relative risk of an event within 24 months after registration among patients who were classified into the high-risk group relative to patients who were classified into the low-risk group was 2.2 (95% confidence interval, 1.41 to 3.51). An up-front GEP biomarker, such as the FL24Cx, rigorously validated in a clinical laboratory and with a clinically relevant turnaround time, could identify and steer enrollment of patients at high risk for early events in clinical trials, thus enabling timely interpretation of such trials and increasing the pace of innovation.
Artificial intelligence accelerates the interpretation of measurable residual B lymphoblastic leukemia by flow cytometry
Abstract Measurable residual disease (MRD) assessment by flow cytometry (FC) plays an essential role in prognosis and therapy escalation of B-cell acute lymphoblastic leukemia (B-ALL). However, the high degree of expertise and manual analysis time required limits the availability of this assay. To overcome this limitation, we developed a data-enhancing artificial intelligence (AI) pipeline that accelerates and simplifies MRD analysis. Unaltered FC files from 171 B-ALL MRD-positive and 89 MRD-negative cases were processed through an AI pipeline trained with 31 expert-gated negative controls. Cluster-informed downsampling reduced FC files from 1.2 million to 155 884 cells per case, on average, (87% cellularity reduction), whereas preserving small MRD populations (median, 100% retention for MRD of <1%) and allowing for true percentage MRD estimates using a correction factor. A deep neural network cell classifier automatically identified normal hematopoietic subsets (macro-averaged F1 score of 0.86); and an AI measure of anomaly discriminated B-ALL from benign mononuclear (area under the curve [AUC] of 0.98) or B-lymphoid cells (AUC of 0.94). Manual analysis of AI-enhanced files was completed in only 1.01 minutes per case, on average (standard deviation of ±0.57); with 100% positive agreement with conventional analysis (for MRD of ≥0.01%), 100% negative agreement, and excellent quantitative correlation (R2 = 0.92). Our cloud-based AI-enhancement solution accelerates B-ALL MRD identification without compromising test performance and has the potential of facilitating B-ALL MRD analysis by more clinical laboratories.
Abstract A148: Increase in colorectal cancer screening among rural and remote Alaska Native peoples using the multi-target stool DNA: A randomized controlled trial
Abstract Background: Alaska Native people are committed to increasing colorectal cancer (CRC) screening to reduce the high rate of CRC in this population. We conducted a community-level cluster randomized controlled trial (RCT) to evaluate the impact of multi-target stool DNA test introduction (mt-sDNA; Cologuard® Exact Sciences, Madison WI) on CRC screening in southwest Alaska Native communities using a mt-sDNA pre-processing lab in Anchorage, Alaska. Methods: A total of 32 rural/remote Alaska Native communities were evenly randomized to either high intensity patient navigation or medium intensity mailed health education compared to 14 communities receiving standard of care (screening recommendations at health care visits). The intervention was conducted from April 2021-February 2023. Outcome measures included CRC screening completion within 12 months and method used (mt-sDNA vs colonoscopy). Results: A total of 2032 Alaska Native participants (mean [SD] age, 56 [8.4] years; 817 women [40%]) were enrolled: 587 participants randomized to usual care, 696 participants randomized to the mailed education arm, and 749 participants randomized to the patient navigation arm. The mt-sDNA test was significantly preferred over colonoscopy by both mailed education participants (68%) and patient navigation participants (71%) (p<0.01). Mt-sDNA was somewhat more preferred by older participants ages 65-75 (79%) vs younger participants ages 45-54 (64%). Both men (71%) and women (68%) significantly preferred mt-sDNA compared with colonoscopy (p<0.01). Overall, 38% of those in the patient navigation intervention completed screening compared with 33% in the mailed education intervention and 10% in the usual care arm. No significant differences were observed in mt-sDNA completion between the patient navigation arm (154 participants [40%]) and the mailed education arm (100 participants [36%]) (p= 0.34). Colonoscopy completion (not including follow-up colonoscopies after abnormal stool tests) was higher in both the patient navigation arm (112 participants [44%]) and the mailed education arm (87 participants [43%]) compared with usual care (38 participants [14%]). Over 61% of kits sent to participants were completed and returned, however, 38% of the kits were not able to be processed, primarily because of mail transit time exceeding the manufacturer’s collection-to-processing limitation of 72 hrs. Conclusions: In this RCT among Alaska Native peoples who were not up to date with CRC screening recommendations, both a mailed education intervention and a patient navigator facilitated intervention were effective at increasing CRC screening compared with usual care. Participants overwhelmingly preferred mt-sDNA for screening but limitations in getting samples to the lab in time for analysis meant that about one-third of mt-sDNA samples were unable to be processed. These findings show that screening outreach and mt-sDNA use can substantially increase CRC screening, but shipping and weather issues remain to be addressed for expanding use of mt-sDNA in rural/remote Alaskan communities. Citation Format: Diana Redwood, Christie Flanagan, Lauren Jeffries, Peter Holck, Danika Bailie, Christine Desnoyers, Joseph Klejka, Lila Rutten, John Kisiel. Increase in colorectal cancer screening among rural and remote Alaska Native peoples using the multi-target stool DNA: A randomized controlled trial [abstract]. In: Proceedings of the 18th AACR Conference on the Science of Cancer Health Disparities; 2025 Sep 18-21; Baltimore, MD. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2025;34(9 Suppl):Abstract nr A148.
Abstract B031: Place, people, and prevention: Contextual drivers of cancer health equity in three US Regions
Abstract Introduction. Understanding the unique social and environmental contexts of communities is essential for designing effective, locally tailored cancer intervention strategies. This study aimed to identify key findings from a comprehensive, multifaceted, cancer-focused needs assessment conducted across three U.S. regions, offering insights into the social, behavioral, and structural barriers that hinder cancer prevention and control, and informing regionally tailored interventions. Methods. This study focused on adult residents within the Mayo Clinic Comprehensive Cancer Center catchment areas, covering 75 counties across Arizona, Florida, and the Midwest (Minnesota, Wisconsin, and Iowa). Data was collected through collaborations with the University of Florida and Mayo Clinic Survey Research Centers, and the “Boots on the Ground" approach involving community health workers. Methods included phone calls, online submissions, and community outreach for an anonymous survey. Validated measures assessed healthcare access, lifestyle, social networks, neighborhood conditions, and discrimination. Descriptive analyses were conducted using valid responses. Non-responses and selections of “Prefer not to answer” were excluded from percentage calculations to ensure interpretive accuracy. Results. Involving 2,681 respondents, the data reveal consistent yet regionally nuanced barriers to cancer prevention. Health insurance coverage was relatively high across all regions—77.6% in Arizona, 85.5% in Florida, and 89.2% in the Midwest. Despite this, access to care remained limited. In Arizona, 22% of respondents could not afford to see a regular doctor, and 41.3% reported delays due to work or transportation. In Florida, 26.6% reported financial barriers to primary care, and nearly half (49.7%) experienced delays in receiving care. The Midwest showed fewer neighborhood safety concerns but still faced high rates of care delays (46.7%) and housing-related issues (42.6%). Neighborhood-level challenges were common. Housing-related problems were reported by 42.9% of respondents in Arizona, 54% in Florida, and 42.6% in the Midwest. Concerns about drug use were highest in Arizona (40%) but present across all regions. Health behavior risks were also widespread: 72–79% of respondents reported sleep difficulties, and many reported low physical activity, suboptimal sun protection, and routine alcohol use. Psychosocial stressors were prominent, with 51–65% of respondents reporting experiences of disrespect and 42–66% feeling unheard by healthcare providers. Conclusions. Despite relatively high rates of insurance coverage, communities across all regions face persistent challenges in accessing timely, affordable, and respectful healthcare. Coupled with neighborhood instability, adverse health behaviors, and psychosocial stressors, these barriers highlight the urgent need for regionally tailored strategies that strengthen both cancer prevention and care. Addressing these interconnected challenges at the community level is essential for promoting more equitable cancer outcomes. Citation Format: Tingyu Zou, Monica L. Albertie, Emelina Asto-Flores, Manisha Salinas, Farhia Omar, Adeline Abbenyi, Noreen Stephenson-Hubert, Nuwanthi Heendeniya, Kathleen Yost, Folakemi T. Odedina. Place, people, and prevention: Contextual drivers of cancer health equity in three US Regions [abstract]. In: Proceedings of the 18th AACR Conference on the Science of Cancer Health Disparities; 2025 Sep 18-21; Baltimore, MD. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2025;34(9 Suppl):Abstract nr B031.
Abstract C138: Influence of demographic factors and knowledge on prostate cancer screening attitude and research participation among Black men
Abstract Introduction: The disproportionate prostate cancer (CaP) burden, which includes high mortality rates, is significantly attributed to the characteristics of the social determinants of the health status of the Black Male population. Designing targeted interventions to address poor health outcomes through a collaborative population-based approach could reduce CaP health disparities. The Community Living Lab (CoLLab) Learning Health System was developed to address ethnically sensitive, targeted approaches that will eliminate CaP health disparities among Black Men. This study assessed demographic factors and Knowledge on prostate cancer screening attitudes and research participation among black men. Method: The collab project employed a social ecological model to design intervention to address barriers to CaP screening and participation in clinical trials. A validated, pretested questionnaire was used to collect information on demographics, CaP knowledge, attitudes, testing, and a history of participation in research and clinical trials. Baseline data were collected from 3 American Legion posts (ALPs). The association between variables was evaluated using the Chi-square test. Results: Participants between 60-69 years were 28.38%, and those aged 40-49 were 12.16%. Most are US-born Black (89.8%), with others identifying as African-born Black (6.76%) and Caribbean-born Black (1.35%). Approximately 43.24% were retired, and 36.49% were working full-time. CaP knowledge was high in 43.9%, moderate in 46.6%, and low in 9.5%. Only 45% of the participants have undergone Prostate-Specific Antigen test, and 24.6% have had a Digital Rectal Examination. Reported past clinical trials and research participation were very low (9.38%). Participants with higher knowledge levels had higher information scores (p < .0001), favored various research settings (p = 0.0008), had favorable attitudes towards CaP testing (p < .0001), and were more likely to participate in research (p = 0.0117). Work status was found to impact CaP Knowledge (p = 0.0117). Participants with stronger beliefs about prostate cancer may have different attitudes toward testing (p = 0.0444). Age did not impact attitudes toward CaP testing (p = 0.4298). Conclusion: The knowledge of CaP was found to impact the attitudes, testing, and disposition of the participants towards research. The baseline data generated from the clinical study will curate insightful evidence that will create innovative approaches to prostate cancer education among males in Black communities. Citation Format: Opeyemi O. Bolajoko, Quincy A. Wimberly, Daniel W. Lee, Collisa Mahin, Michelle Fudge, Shaina E. Flynn, Arnold J. Merriweather, Trevanne R. Matthews Hew, Monica L. Albertie, Dery Greene, Wayne Ford, Toshiko Moultrie, Floyd B. Willis, Folakemi T. Odedina. Influence of demographic factors and knowledge on prostate cancer screening attitude and research participation among Black men [abstract]. In: Proceedings of the 18th AACR Conference on the Science of Cancer Health Disparities; 2025 Sep 18-21; Baltimore, MD. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2025;34(9 Suppl):Abstract nr C138.