Expert Physicians Research Articles

Predictive Value of Pre-Infusion FDG‐PET/CT Parameters for Clinical Outcomes and Toxicities in Relapsed/Refractory (R/R) Large B-Cell Lymphoma (LBCL) Following CAR-T Cell Therapy

Introduction: CD19-targeting chimeric antigen receptor (CAR) T-cell therapy has demonstrated significant efficacy in patients with relapsed/refractory (R/R) non-Hodgkin lymphoma but is associated with serious adverse effects such CRS and ICANS. Pre-infusion disease burden has been investigated as a potential predictor of clinical outcome and toxicity. FDG-PET/CT derived metabolic tumor volume (MTV) and total lesion glycolysis (TLG) are quantitative indicators of this baseline tumor load. We aim to assess the predictive value of pre-CART PET metrics on outcome and toxicity in R/R large B-Cell Lymphoma (LBCL) following CAR-T cell therapy. Method: We conducted a retrospective single site study of 25 consecutive patients with R/R LBCL treated with CAR-T therapy in our institution between May 2020 and October 2022, All patients underwent FDG-PET/CT imaging pre-CAR-T infusion (PET-0) and at day 30 post infusion (PET-1). The images were reviewed by two expert lymphoma PET physicians. Metabolic parameters including MTV, TLG (both calculated using a relative threshold of 40%), SUVmax and Deauville score were calculated. PET-1 was classified as complete metabolic response (CMR, Deauville score 1-3) or non-CMR (Deauville score 4-5) as per the Lugano classification. Results: 25 patients were analyzed , Based on the median MTV (74,6 mL), TLG (379,4), SUVmax (11,8) values, patients were divided into two groups “low” and “high”. The median follow-up was 12 months. Patients with low pre-infusion MTV and TLG had significantly superior day 30 CMR rates (84,6% vs 25%, p= 0.001 & 80% vs 20%, p = 0.001 respectively). Pre-infusion SUVmax ( p = 0.29 ) and Deauville score ( p = 0.103), however, did not predict response to treatment. In our population, pre-infusion PET parameters were not associated with CRS or ICANS . Patients with low pre-infusion MTV and TLG had significantly superior PFS (MTV: p=0.02; TLG: p=0.019) and OS (MTV: p=0.017; TLG: p=0.05), Pre-infusion SUVmax had no statistical significant impact on OS ( p = 0.61) or PFS ( p = 0.37) Conclusion: In our study, pre-infusion quantitative PET metrics, TLG and MTV, predict response and outcomes of CART therapy in R/R LBCL. Our results highlight the importance of adequate disease control before CAR-T administration and the value of pre-infusion FDG-PET/CT as a predictor of outcomes in R/R LBCL patients treated by CAR-T-therapy.

Reducing Preventable Bleeding Visits and Institutional Factor Costs for Patients with Bleeding Disorders with a Prospective Quality Improvement Project

Providing optimal care to patients with bleeding disorders requires specialized physician, laboratory, pharmacy, nursing, and institutional resources best achieved at Hemophilia Treatment Centers (HTCs). Current systems for managing the care of a resource intensive patient population may be suboptimal. We observed frequent bleeding related hospitalizations and emergency room (ER) visits in patients with bleeding disorders treated at our HTC that could have been potentially prevented by changes in outpatient care delivery. We initiated a prospective quality improvement (QI) project to improve care delivery and to reduce preventable bleeding related hospitalizations and ER visits for adult patients treated at our HTC. The interventions targeted patient-level and systems-level factors contributing to preventable visits, and barriers to care in our patient population. We compared preventable bleeding related hospitalizations and ER visits, institutional factor cost, use of prophylaxis in severe hemophilia, primary care physician attainment, and patient electronic portal enrollment before and after the QI intervention. In our population of adult patients with bleeding disorders (N=88), the intervention significantly reduced the primary outcome of preventable bleeding hospitalization or ER visit (P=0.0082, Figure), and reduced such visits by 85.4%, meeting the pre-specified criteria for success. Institutional factor acquisition costs to treat preventable bleeding were reduced by 94.5%, from $11,800 to $640 per patient per year; a projected savings of $982,088 yearly. Use of factor prophylaxis in patients with severe hemophilia increased from 58.8% to 100%, attainment of a primary care physician increased from 69.4% to 86.4%, and use of the electronic patient portal increased from 53.2% to 83%; all secondary outcomes met pre-specified criteria of success and were statistically significant. HTC clinic visit attendance was low at 55.2%. The majority of patients (71.6%) had at least one outpatient urgent episode (mean 0.72 episode per year), and 93% had non-urgent management (mean 9.3 episodes per year) occurring outside of a clinic visit. These data demonstrate that concerted outpatient management of patients with bleeding disorders with a quality improvement framework can reduce preventable hospitalization and ER visits. Decreased inpatient utilization resulted in substantial reductions in factor costs at the health system level. The clinical effort required to achieve these results was substantial, and required frequent provider engagement outside of regular clinic visits. The HTC can improve patient outcomes and reduce institutional costs from preventable bleeding episodes by supporting a collaborative quality improvement process led by physician, nursing, and laboratory coagulation experts.

Comparative Analysis of Bone Marrow Microbiome and Proteome Differences between Essential Thrombocythaemia and Prefibrotic Primary Myelofibrosis

Background Philadelphia chromosome-negative myeloproliferative neoplasms (MPNs) are hematologic malignancies with a hallmark feature of chronic inflammation. MPNs mainly include polycythaemia vera (PV), essential thrombocythaemia (ET), primary myelofibrosis (PMF) and prefibrotic primary myelofibrosis (pre-PMF). In contrast to ET, an accurate morphological diagnosis of pre-PMF is a key issue due to its worse prognosis. However, it is difficult to accurately diagnose pre-PMF, mainly relying on pathological diagnosis. The heavy reliance on morphology for accurate diagnosis poses a challenge to the professional abilities of physicians. The uneven expertise of physicians can lead to improper diagnosis and delay patient treatment. Therefore, accurate diagnosis of these two disease subtypes is of great importance. It is increasingly recognized that the bone marrow microenvironment, including inflammation and metabolic imbalances, influences disease progression. So far, there have been few studies focusing on the combined analysis of the bone marrow microbiome and proteome in MPN patients. We aim to describe the changes of bone marrow microbiome and proteome and their association with pathogenesis in ET and pre-PMF patients Method We analyzed the formalin-fixed, paraffin embedded (FFPE) tissue slide of bone marrow from six patients with untreated pre-PMF (median age: 64 years, m/f=4/2) and ET (n=7, median age: 53 years, m/f=2/5). Proteomic analysis of FFPE tissue slides of bone marrow from 13 patients was performed using 4D-direct DIA. The bone marrow microbiota was analyzed by using 2bRAD-M sequencing technology. Result The clinical and laboratory characteristics of ET and pre-PMF patients are summarized in Table 1. The JAK2 V617F mutations were detected in all patients, and the mutations were more frequent in patients with pre-PMF than in those with ET (17.9% vs. 55.8%, p=0.001). All patients in the ET group are at MF grade 0, while all in the pre-PMF group are at MF grade 1. The lactate dehydrogenase in pre-PMF group patients is higher than in ET group patients, and the difference is statistically significant ( p=0.007). In addition, the neutrophil-to-lymphocyte ratio (NLR), an inflammation-related indicator, is significantly increased in the pre-PMF group. The neutrophil-to-lymphocyte ratio in the pre-PMF group is significantly higher than in the ET group, which is statistically significant ( p=0.015). High density lipoprotein cholesterol (HDL) in the pre-PMF group is significantly reduced, and the difference is statistically significant ( p<0.001). This suggests a difference in lipid metabolism between pre-PMF and ET. Figure 1A shows a significant correlation between NLR and HDL, with a statistical difference ( p= 0.013). Proteomics analysis identified a total of 177 differential proteins in the bone marrow of ET and pre-PMF, of which 108 were up-regulated and 69 were down-regulated in the pre-PMF group (Figure 1B and 1C). Through functional enrichment analysis, we found that the differentially expressed proteins upregulated in the pre-PMF group (such as: CXCR4, CXCR2, GBP5, etc.) are mainly enriched in immune and inflammation-related GO (Gene Ontology) terms, such as immune response (GO:000695). The differentially expressed proteins downregulated in the pre-PMF group (such as APOA4, FABP4, etc.) are mainly enriched in lipid metabolism-related GO terms, such as lipoprotein metabolic process (GO: 0042157), and positive regulation of fatty acid biosynthetic process (GO: 0045723). Association analysis revealed a significant relationship between differential proteins and clinical markers (Figure 1D). We identified 223 differential microbes in the bone marrow, which were classified into 88 genera. The α diversity of the microbiota in bone marrow calculated by Chao1 showed that there was significant difference in species diversity between the two groups (Figure 1E). Microbiome composition showed statistically significant differences between pre-PMF and ET ( p<0.001, R 2=0.302, PERMANOVA). Microbial diversity and richness was higher in pre-PMF, and reaching statistical significance in this small cohort (Figure 1F). Conclusion We have shown that there are multi-level differences, including the proteome and microbiome, between ET and pre-PMF patients.This could potentially provide new insights for distinguishing and treating these two disease subtypes.

Drepadom - a Homecare Protocol for Vaso-Occlusive Crisis for Sickle Cell Disease Patients

INTRODUCTION Vaso-occlusive crisis (VOC) is first cause of emergency room (ER) admissions and hospitalizations for sickle-cell disease (SCD) patients. Acute chest syndrome (ACS) is a complication that can occur during VOC and prolong hospitalization, and is the main cause of death in SCD patients. The PRESEV score (Bartolucci - eBioMed 2016), established by our team, allows physicians to evaluate the risk of developing ACS and is calculated upon arrival to the ER and has been validated by an international multicenter study in five different countries (PRESEV 2 - ASH 2020). During the Covid-19 pandemic, we set up a temporary system to monitor VOC patients at home using a telephone hotline and homecare services. The success of this system led to the establishment of DREPADOM, a home-care protocol for VOC management for patients with low risk of developing ACS, or who can be discharged early from the hospital, as standard care (DREPADOM - ASH 2021). Since 2021, Drepadom has been successfully running in seven hospitals in the Parisian region. DESCRIPTION OF SETTING Drepadom has two inclusion modes: 1/ Patients arriving at the ER for VOC with a PRESEV score ≤ 6 and not presenting any exclusion criteria; 2/ patients that are discharged early following hospitalization for VOC and/or ACS (after at least 3 days of hospitalization for patients with PRESEV > 6 upon ER arrival). The Drepadom homecare protocol is activated by a coordinator, who coordinates delivery of PO opioids and parenteral treatments, and a home oxygen concentrator, to the patients' home within four hours. Drepadom uses an online platform and mobile app (Link4Life), where the patient's medical record is shared to the homecare nurse who will perform house calls twice a day. During each visit, the nurse enters the patient's vitals into the app, which are monitored by an on-call SCD expert physician. The online platform also hosts a secure messaging service where nurses, doctors and coordinators can communicate in real-time regarding patient status and care, as well as an automatic alert system that is triggered when vitals fall outside of a predetermined range set by SCD experts. Patient safety is further ensured with a daily calls to the patient using a decision-making algorithm supervised by SCD experts, who arbitrate according to the evolution of the situation (stopping/continuing outpatient follow-up vs. rehospitalization). RESULTS Over a 30-month period we had 209 inclusions for 170 patients into DREPADOM. Of these, 105 were inclusions from the ER and 104 were early discharges (ED). SCD genotype distribution was 83.2% SS, 10.5% SC, and 5.8% S-bThalassemia and female to male ratio was 128:81. In both ER and ED groups, median age was 31 [27-40], 74% of patients were under hydroxyurea treatment and median homecare follow-up duration was 3 [2-4] days. No significant differences in population characteristics were found between the ER and ED groups. In the ER group, 96/105 patients (91.4%) were included with PRESEV ≤ 6. 9/105 (8.6%) patients were included with a PRESEV score >6 only in 2021 during the Covid-19 pandemic, when the ER was saturated. In the PRESEV ≤ 6 group, 9/96 (9.4%) patients were rehospitalized: 1 for ACS (PRESEV = 4), 4 for uncontrolled VOC pain, 2 for chest pain, 2 for fever/infection unrelated to VOC. In the PRESEV >6 group, 3/9 (33%) were hospitalized, one for ACS (PRESEV = 8), one for uncontrolled VOC pain and one for issues with veinous access. In the ED group, patients were hospitalized for a median of 4.5 [3-7] days before being discharged. 27 patients (26%) were included in a post-ACS context. 48/104 (46.2%) had a PRESEV ≤6, of which only 2/48 (4.2%) were rehospitalized for uncontrolled VOC pain. Of the 56/104 (53.8%) patients included with a PRESEV > 6, 8/56 (14.3%) were rehospitalized: 1 for ACS, 3 for uncontrolled VOC pain, 1 for chest pain, 1 for fever/infection unrelated to VOC, 1 for issues with veinous access, and 1 for morphine withdrawal symptoms. For both ER and ED groups, a total of 22/209 patients (11%) required rehospitalization (Figure 1), and no deaths or ICU were reported in either group. PERSPECTIVES Our results confirm that Drepadom is a successful and safe tool to reduce hospital stay for SCD patients.

Inpatient Use of Extended-Release (ER) Opioids for Vaso-Occlusive Crisis Associated with Increased Length of Stay for ER Opioid-Naïve Patients

Background: The intermittent acute pain syndrome associated with sickle cell disease (SCD) is a poorly understood phenomenon with complex pathophysiology. Given the broad spectrum of analgesic interventions, including immediate-release (IR) and extended-release (ER) opioids, healthcare providers are challenged with the task of safely and effectively escalating inpatient opioid regimens, especially in the setting of limited guidelines outlining the combination of different opioid formulations. Despite SCD being the most common genetic disorder in the United States, more research is required to improve physician expertise with appropriate and timely pain management of vaso-occlusive crisis. Methods: This study was a retrospective chart review (n=200) of patients with known SCD requiring hospitalization for acute pain secondary to vaso-occlusive crisis without an alternative etiology of pain symptoms. The review included a patient population aged 18-67 years old with 43.5% males and 56.5% females hospitalized from 2/1/2022 to 5/6/2022. The primary outcome measure was hospital length of stay (LOS) for adequate pain control compared against varying factors of inpatient opioid-prescribing practices, including initiation or continuation (if prescribed within 6 months prior to admission encounter per PDMP review) of an ER opioid formulation, degree of morphine milligram equivalents (MME) escalation from outpatient regimen, utilization of a sickle cell analgesic order set, and initiation of patient-controlled analgesia pump (PCA). A secondary outcome measure investigated was the quantitative comparison of highest daily prescribed MME in relation to the aforementioned inpatient variables. Data analysis was completed using Wilcoxon Rank Summary analysis, with a significant p-value set to < 0.05. Results: Among the primary outcome analysis, the only significant factor affecting LOS was the introduction of an ER opioid. The results show that within the patient cohort naïve to an outpatient ER opioid (n=105), ER opioid-incorporated inpatient analgesic regimens resulted in a significantly increased mean LOS compared to IR opioid monotherapy regimens, with an average of 8.8 inpatient days as compared to 5.3 days, respectively (p-value < 0.001). For the patients prescribed an outpatient ER opioid at baseline (n=95), the continuation of an inpatient ER opioid did not result in a significant change in mean LOS, calculated as 6.7 days vs. 6.0 days in the continued and non-continued groups, respectively (p-value = 0.716). Among the secondary outcome measures, utilization of a sickle cell order set and PCA did not significantly affect degree of MME escalation. Conclusions: Limited evidence-based recommendations for the treatment of acute pain episodes in SCD has resulted in widely varying opioid regimens prescribed by inpatient providers and has created clinical challenges regarding the optimization of pain management. This study investigated inpatient opioid regimens utilized in treatment of vaso-occlusive crisis and found that the initiation of ER opioids resulted in increased length of stay for ER opioid-naïve patients as compared to treatment with IR opioids alone. These results question the benefit of inpatient ER opioids in relation to the risk of opioid dependence and MME escalation.

772. Learning from our patients: An exploratory study to inform the development of a case tracking dashboard for subspecialty fellows

Abstract Background Educators identify case tracking (following-up prior patient encounters) as a valued practice to cultivate physician expertise. Much of the existing research has focused on the practices of general medicine providers, where challenges include insufficient time and difficulty tracking patients in the electronic health record (EHR). Subspecialty fellows constitute a group of advanced learners who see a high volume of patients and need to quickly become subject matter experts, yet little is known about their particular needs and motivations around case tracking. In this study, we performed a needs assessment with fellows to inform the design of an EHR-based case tracking dashboard that can facilitate learning from prior patient encounters. Methods We conducted semi-structured interviews with 22 medicine subspecialty fellows (including 4 Infectious Diseases [ID] fellows) at a single academic medical center that explored participants’ motivations, preferences, and practices around case tracking. Interview transcripts were coded and analyzed with directed qualitative content analysis. Results of this analysis will be used to inform the design of a subspecialist EHR-based case tracking dashboard. Results Participants were motivated to engage in case tracking, with curiosity regarding patient outcomes as a strong motivator. Although some attempted to track all consults, most focused on subsets of patients, feeling that not all encounters were of equal educational value. Participants highlighted particular results, specialty-specific notes, and major outcomes as the most useful information. ID fellows specifically were interested in post-discharge ID clinic encounters, send-out tests, and outcomes of their management decisions. Despite widespread motivation to engage in case tracking, manually adding patients to lists and remembering to follow up poses logistical and time barriers. Conclusion An ideal EHR case-tracking dashboard would automatically keep a list of patients cared for by fellows and provide easy access to own-specialty notes, specific results of interest, and information on the global clinical course. The ability to highlight specific cases anticipated to have more educational value would also be of benefit. Disclosures All Authors: No reported disclosures

Artificial Intelligence–Aided Selection of Needle Pathways: Proof-of-Concept in Percutaneous Lung Biopsies

PurposeTo evaluate the concordance between lung biopsy puncture pathways determined by artificial intelligence (AI) and those determined by expert physicians. Materials and MethodsAn AI algorithm was created to choose optimal lung biopsy pathways based on segmented thoracic anatomy and emphysema in volumetric lung computed tomography (CT) scans combined with rules derived from the medical literature. The algorithm was validated using pathways generated from CT scans of randomly selected patients (n = 48) who had received percutaneous lung biopsies and had noncontrast CT scans of 1.25-mm thickness available in picture archiving and communication system (PACS) (n = 28, mean age, 68.4 years ± 9.2; 12 women, 16 men). The algorithm generated 5 potential pathways per scan, including the computer-selected best pathway and 4 random pathways (n = 140). Four experienced physicians rated each pathway on a 1–5 scale, where scores of 1–3 were considered safe and 4–5 were considered unsafe. Concordance between computer and physician ratings was assessed using Cohen’s κ. ResultsThe algorithm ratings were statistically equivalent to the physician ratings (safe vs unsafe: κ¯=0.73; ordinal scale: κ¯=0.62). The computer and physician ratings were identical in 57.9% (81/140) of cases and differed by a median of 0 points. All least-cost “best” pathways generated by the algorithm were considered safe by both computer and physicians (28/28) and were judged by physicians to be ideal or near ideal. ConclusionsAI-generated lung biopsy puncture paths were concordant with expert physician reviewers and considered safe. A prospective comparison between computer- and physician-selected puncture paths appears indicated in addition to expansion to other anatomic locations and procedures.

Cost-effectiveness of implant movement analysis in aseptic loosening after hip replacement: a health-economic model

ObjectiveTo investigate the cost-effectiveness of using Implant Movement Analysis (IMA) to follow up suspected aseptic loosening when the diagnosis after an initial X-ray is not conclusive, compared with a diagnostic pathway with X-ray follow-up.MethodsA health-economic model in the form of a decision tree was developed using quality-adjusted life years (QALY) from the literature, cost-per-patient data from a university hospital and the probabilities of different events from expert physicians’ opinions. The base case incremental cost-effectiveness ratio (ICER) was compared with established willingness-to-pay thresholds and sensitivity analyses were performed to account for assumptions and uncertainty.ResultsThe base case ICER indicated that the IMA pathway was cost effective (SEK 99,681, compared with the SEK 500,000 threshold). In the sensitivity analysis, the IMA pathway remained cost effective during most changes in parameters. ICERs above the threshold value occurred in cases where a larger or smaller proportion of people receive immediate surgery.ConclusionA diagnostic pathway using IMA after an inconclusive X-ray for suspected aseptic loosening was cost effective compared with a pathway with X-ray follow-up.

Wavelet transform and deep learning-based obstructive sleep apnea detection from single-lead ECG signals.

Sleep apnea is a common sleep disorder. Traditional testing and diagnosis heavily rely on the expertise of physicians, as well as analysis and statistical interpretation of extensive sleep testing data, resulting in time-consuming and labor-intensive processes. To address the problems of complex feature extraction, data imbalance, and low model capacity, we proposed an automatic sleep apnea classification model (CA-EfficientNet) based on the wavelet transform, a lightweight neural network, and a coordinated attention mechanism. The signal is converted into a time-frequency image by wavelet transform and put into the proposed model for classification. The effects of input time window, wavelet transform type and data balancing on the classification performance are considered, and a cost-sensitive algorithm is introduced to more accurately distinguish between normal and abnormal breathing events. PhysioNet apnea ECG database was used for training and evaluation. The 3-min Frequency B-Spline wavelets transform of ECG signal was carried out, and Dice Loss was used to train the classification model of sleep breathing. The classification accuracy was 93.44%, sensitivity was 88.9%, specificity was 96.2% and most indexes were better than other related work.

Nixon S. See, MD (1962-2023)

While I am honored to be chosen among his many friends to pay tribute to him, I am struck by ambivalence, as I mourn just as deeply for Nixon’s passing. Nevertheless, as this is likely my last opportunity to praise this great man, I aim to give a good accounting of his life and times.
 Nix and I first met as applicants of the residency program in ORL-HNS at the Ospital ng Maynila Medical Center (OMMC) in September 1989. Subsequently, he established his practice in his hometown of Calapan, Oriental Mindoro. He was briefly affiliated with the PSO-HNS Southern Tagalog Chapter as a charter member but was forced to relinquish this and a flourishing practice and return to Manila to care for his wife. He joined the Fatima University Medical Center and College of Medicine, eventually becoming chairman of the Department of ORL-HNS. He rejoined OMMC as consultant in 2010.
 Having been his close friend for more than three decades, I have known him to be a peaceful man. Indeed, he abhorred life’s unnecessary complexities, wanting only a simple, quiet life, content in the knowledge that he has filled the cups of those whose lives he had touched.
 I use PEACEFUL, not only in a literal sense, but also as an acronym that represents his qualities I have learned to love, as well as hate. Nix was a Principled man, to the point of being stubborn. His moral compass was always straight and true, unwavering, and without any gray areas. He was also a Practical man, choosing to be frugal for himself, yet generous to his family. He was Exacting, meticulously insisting on accomplishing tasks in a specific, almost textbook-like manner. “So-so” was not in his vocabulary. An Amiable individual, he was never aloof and was ready to greet anyone with a genuine smile. He chose to live Cleanly, leading a life not only free of vices, but one pursued with healthy physical activity. Being meticulously principled was not without its drawbacks. He was Cranky. During a typical cantankerous mood, he would wipe his face with his palm, as if to “erase” it. This was followed by a long, deep snort. Woe to the junior resident who would earn his ire! I would remember him say: “though I’m a specialist, I don’t think I will ever become filthy rich.” He was a man of charity, an Expert physician: competent, conscientious, caring, and humane. He was a devout Family man, one whose decisions, without exception, revolved around the welfare of his loved ones. While he reached impressive heights professionally, he remained grounded and Unassuming, a true epitome of humility. Lastly, he was a Loyal friend to the end.
 I have lost a brother. Yet, during his final days, having witnessed how his illness ravaged his physical form relentlessly, mercilessly, I take solace in the knowledge that he is now in a place of eternal serenity, free of doubt, fear, and despair. I once read: God is indeed wise because He placed death at the end of life and not at its beginning. Truer words have never been said, for it allowed me to experience the richness of life with Nix. To my friend, I say: “Till we meet again.”

Heart sound classification based on bispectrum features and Vision Transformer mode

In regions with limited resources and moderate incomes, the relentless spectre of cardiovascular diseases (CVDs) continues to loom large. Amidst this challenge, the precise classification of heart sounds is emerging as a pivotal linchpin in the realm of early CVD diagnosis and intervention. Manual heart sound auscultation efficacy remains tethered to the expertise of physicians, but the tides are shifting. With deep learning algorithms, heart sound classification reaches new heights. In this paradigm-shifting article, we unveil an ingenious model fortified by bispectrum-inspired feature extraction and the cutting-edge prowess of the Vision Transformer (ViT) model. This model spearheads the binary classification of heart sounds, labelling them as either 'normal' or 'abnormal.' Our model uses data from the PhysioNet Challenge 2022 database, which contains 3163 data points from 942 patients. The model showcases an adept classification process with a remarkable consistency, notably holding its own when distinguishing between heart sounds of pregnant and nonpregnant patients. Moreover, we dare to challenge the status quo. This article boldly pits the performance of our model against that of seasoned cardiologists. Our model emerges as the triumphant frontrunner, eclipsing the proficiency of even the most seasoned cardiologists. In a world where health resources are unevenly distributed, this pioneering work offers a beacon of hope, unlocking a future where advanced algorithms not only match but surpass human expertise. Step into the forefront of transformative cardiac care—where artificial intelligence becomes the key to unravelling the secrets of heart sounds.

Leveraging Large Language Models for Decision Support in Personalized Oncology

Clinical interpretation of complex biomarkers for precision oncology currently requires manual investigations of previous studies and databases. Conversational large language models (LLMs) might be beneficial as automated tools for assisting clinical decision-making. To assess performance and define their role using 4 recent LLMs as support tools for precision oncology. This diagnostic study examined 10 fictional cases of patients with advanced cancer with genetic alterations. Each case was submitted to 4 different LLMs (ChatGPT, Galactica, Perplexity, and BioMedLM) and 1 expert physician to identify personalized treatment options in 2023. Treatment options were masked and presented to a molecular tumor board (MTB), whose members rated the likelihood of a treatment option coming from an LLM on a scale from 0 to 10 (0, extremely unlikely; 10, extremely likely) and decided whether the treatment option was clinically useful. Number of treatment options, precision, recall, F1 score of LLMs compared with human experts, recognizability, and usefulness of recommendations. For 10 fictional cancer patients (4 with lung cancer, 6 with other; median [IQR] 3.5 [3.0-4.8] molecular alterations per patient), a median (IQR) number of 4.0 (4.0-4.0) compared with 3.0 (3.0-5.0), 7.5 (4.3-9.8), 11.5 (7.8-13.0), and 13.0 (11.3-21.5) treatment options each was identified by the human expert and 4 LLMs, respectively. When considering the expert as a criterion standard, LLM-proposed treatment options reached F1 scores of 0.04, 0.17, 0.14, and 0.19 across all patients combined. Combining treatment options from different LLMs allowed a precision of 0.29 and a recall of 0.29 for an F1 score of 0.29. LLM-generated treatment options were recognized as AI-generated with a median (IQR) 7.5 (5.3-9.0) points in contrast to 2.0 (1.0-3.0) points for manually annotated cases. A crucial reason for identifying AI-generated treatment options was insufficient accompanying evidence. For each patient, at least 1 LLM generated a treatment option that was considered helpful by MTB members. Two unique useful treatment options (including 1 unique treatment strategy) were identified only by LLM. In this diagnostic study, treatment options of LLMs in precision oncology did not reach the quality and credibility of human experts; however, they generated helpful ideas that might have complemented established procedures. Considering technological progress, LLMs could play an increasingly important role in assisting with screening and selecting relevant biomedical literature to support evidence-based, personalized treatment decisions.

Implementation strategies for large scale quality improvement initiatives in primary care settings: a qualitative assessment

Background and objectivesThe EvidenceNOW: Advancing Heart Health in Primary Care was designed to assist primary care practices in the US in implementing evidence-based practices in cardiovascular care and building capacity for quality improvement. EvidenceNOW, NCT03054090, was registered with ClinicalTrials.gov on 15/02/2017. The goals of this study were to gain a comprehensive understanding of perspectives from research participants and research team members on the value of implementation strategies and factors that influenced the EvidenceNOW initiative in Virginia.MethodsIn 2018, we conducted 25 focus groups with clinicians and staff at participating practices, including 80 physicians, advanced practice clinicians, practice managers and other practice staff. We also conducted face-to-face and telephone interviews with 22 research team members, including lead investigators, practice facilitators, physician expert consultants, and evaluators. We used the integrated-Promoting Action on Research Implementation in the Health Services (i-PARIHS) framework in our qualitative data analysis and organization of themes.ResultsImplementation strategies valued by both practice representatives and research team members included the kick-off event, on-site practice facilitation, and interaction with physician expert consultants. Remote practice facilitation and web-based tools were used less frequently. Contextual factors that influence quality improvement efforts include leadership support, access to resources, previous quality improvement experience, and practice ownership type (independent compared to health system owned). Many clinicians and staff were overwhelmed by day-to-day activities and experience initiative fatigue, which hindered their ability to fully participate in the EvidenceNOW initiative.ConclusionsThis study provides details on how the practice environment plays an essential role in the implementation of evidence-based practices in primary care. Future efforts to improve quality in primary care practices should consider the context and environment of individual practices, with targeted implementation strategies to meet the needs of independent and health system owned practices. Future efforts to improve quality in primary care practices require strategies to address initiative fatigue among clinicians and practice staff. External support for building capacity for quality improvement could help primary care practices implement and sustain evidence-based practices and improve quality of care.Trial registrationThis project was registered with ClinicalTrials.gov on 15/02/2017 and the identifier is NCT03054090.

Application of the Face2Gene tool in an Italian dysmorphological pediatric clinic: Retrospective validation and future perspectives.

Neurodevelopmental disorders exhibit recurrent facial features that can suggest the genetic diagnosis at a glance, but recognizing subtle dysmorphisms is a specialized skill that requires very long training. Face2Gene (FDNA Inc) is an innovative computer-aided phenotyping tool that analyses patient's portraits and suggests 30 candidate syndromes with similar morphology in a prioritized list. We hypothesized that the software could support even expert physicians in the diagnostic workup of genetic conditions. In this study, we assessed the performance of Face2Gene in an Italian dysmorphological pediatrics clinic. We uploaded two-dimensional face pictures of 145 children affected by genetic conditions with typical phenotypic traits. All diagnoses were previously confirmed by cytogenetic or molecular tests. Overall, the software's differential included the correct syndrome in most cases (98%). We evaluated the efficiency of the algorithm even considering the rareness of the genetic conditions. All "common" diagnoses were correctly identified, most of them with high diagnostic accuracy (93% in top-3 matches). Finally, the performance for the most common pediatric syndromes was calculated. Face2Gene performed well even for ultra-rare genetic conditions (75% within top-3 matches and 83% within top-10 matches). Expert geneticists maybe do not need computer support to recognize common syndromes, but our results prove that the tool can be useful not only for general pediatricians but also in dysmorphological clinics for ultra-rare genetic conditions.

Evaluating the Performance of Large Language Models in Hematopoietic Stem Cell Transplantation Decision Making

Introduction: Large Language Models (LLMs) are a form of Artificial Intelligence (AI), by identifying patterns and connections within data, they can predict the most likely words or phrases in specific contexts. Previous studies have indicated that GPT (Generative Pre-trained Transformer; OpenAI) performs well in answering single-choice clinical questions. However, its performance seems to be less satisfactory when dealing with multiple-choice questions and more intricate clinical cases (Cosima et al. 2023 EAO; Cascella et al. 2023 J Med Syst). Notably, no study has evaluated LLMs responses in the context of Transplantation Decision Making, a complex process heavily reliant on physician expertise. Additionally, most studies focused solely on GPT's performance, without considering other competitive LLMs like Llama-2 or VertexAI. Our study aims to assess the performance of LLMs in the domain of hematopoietic stem cell transplantation. Methods: We modified and anonymized the clinical histories of six hematological patients. An experienced hematologist reviewed and validated these modified clinical histories, which included demographic data, past medical history, hematology disease features (genetic data and MRD when available), treatment responses, adverse events from previous therapies, and potential donor information (related/unrelated, HLA, CMV status). We presented these clinical cases to six experienced bone marrow transplant physicians from two major JACIE accredited hospitals and 11 hematology residents from the University Milano-Bicocca. LLMs employed for the analysis were: GPT-4, VertexAI Palm 2, Llama-2 13b and 70b. LLMs were configured with different temperature settings to control token selection randomness, always maintaining low levels for more deterministic responses. A triple-blinded survey was conducted using Typeform, where both senior hematologists and residents provided anonymized responses with personal tokens. The senior hematologists, residents, and LLMs testers were unaware of the responses provided by the other groups. We calculated Fleiss K (K) and overall percentage of agreement (OA) between residents and LLMs, considering the consensus answer (CoA) among experts as the most frequent response. Subsequently, OA and K values for both residents and LLMs were compared using T- or Mann-Whitney tests with Graphpad v 10.0.1. Results: The results showed perfect agreement among experts in patient transplant eligibility assessment (K=1.0) and substantial agreement in the choice of donors and conditioning regimens (K=0.62 for both questions). Fair agreement was observed in Transplant Related Mortality (TRM) estimation (K=0.22). The median OA and K value between residents and the CoA of experts were 76.5% (range 52.9-88.2%) and 0.61 (range 0.4-0.8), respectively. The median OA and K value between LLMs answers and experts were 58.8% (range 47-71%) and 0.45 (range 0.3-0.61), respectively. The mean OA and K value of residents were significantly higher compared to LLMs (p=0.02). Specifically, residents showed higher median OA and K values in patient eligibility assessment (median OA 100 vs. 83% and K 1 vs. 0.78; p=0.01). However, there was no significant difference in median K for donor choice (0.56 vs. 0.56), conditioning regimen (0.67 vs. 0.33), and TRM evaluation (0.33 vs. 0) (Table 1). The median K values of GPT-4, Palm-2, Llama2-13b, and Llama2-70b were 0.49, 0.53, 0.33, and 0.53 respectively (Figure 1). Conclusion: Our study sheds light on the potential and limitations of LLMs in complex hematopoietic stem cell transplantation decision-making. While LLMs showed promising results with a median OA of 59%, residents demonstrated superior performance. LLMS displayed good performances in patients' eligibility and donor choice but showed shortcomings in conditioning regimens and TRM evaluation. Not using a rating scale from experts when evaluating LLMs responses aimed to avoid potential bias. However, it is important to note that the consensus answer, even though it was the most frequent, does not necessarily imply that other responses provided by the experts were incorrect. Therefore, the lower consensus among the experts in TRM evaluation, possibly due to the challenge of precisely calculating TRM in a survey-based evaluation, should also lead to a cautious approach when evaluating residents and LLMs answers in this setting.

AHOD2131: A Randomized Phase 3 Response-Adapted Trial Comparing Standard Therapy with Immuno-Oncology Therapy for Children and Adults with Newly Diagnosed Stage I and II Classic Hodgkin Lymphoma

Background: Chemotherapy in combination with radiotherapy (RT) is the standard for early-stage (ES) Hodgkin lymphoma (cHL). Despite high cure rates (12-year overall survival (OS) >93%), there is room to improve short and long-term outcomes for children and adults with ES cHL. Incorporation of immunotherapy (IO) into front line treatment of ES cHL is an opportunity to both improve progression-free survival (PFS) and maintain OS, while minimizing long-term morbidity and mortality by reducing exposure to RT and high-dose chemotherapy. Study Design and Methods: HL leaders of the NCTN groups, expert HL researchers and physicians, and representatives from the Cancer Therapy Evaluation Program (CTEP) formed a working group in April 2020 with the goals of harmonizing treatment approaches for ES cHL across pediatric and adult patients and reaching consensus regarding the optimal study design for incorporating IO therapy into frontline treatment. Study champions identified from each of the North American cooperative groups [Children's Oncology Group (COG), SWOG, ECOG-ACRIN, Alliance, NRG] and experts in imaging, radiation oncology, lymphoma biology and patient-reported outcomes (PROs) were included. The resulting COG-led clinical trial, AHOD2131, represents the largest ES cHL trial in the history of North American cooperative groups and the first to enroll patients across the age continuum. AHOD2131 (NCT05675410; Figure) is a randomized, phase 3 trial enrolling patients ages 5 to 60 years with newly diagnosed, previously untreated stage I and II cHL and will investigate the addition of the CD30-antibody drug conjugate brentuximab vedotin (Bv) with PD-1 blockade (nivolumab) compared to standard chemotherapy +/- RT. Since the April 2023 activation, 97 North American sites have activated the trial, and 21 participants have enrolled. Target enrollment is 1875 patients over 5 years of accrual, for an estimated 1782 eligible and evaluable patients. The primary objective is to compare the 3-year PFS of patients with ES cHL treated through a response-adapted, superiority design with either standard therapy or with an IO approach (BV + nivolumab). All patients will be stratified as favorable or unfavorable based on disease risk features at enrollment. Based on initial response assessment by PET/CT (via central review) after 2 cycles of ABVD, patients were classified to PET2 positive (SER, defined as 5 Point Score 4 or 5, ~15% of patients) and PET negative (RER). Those who are PET positive will receive involved site RT (ISRT). The SER and RER patients will be randomized to standard chemotherapy vs. IO therapy respectively. In addition to the primary outcome of PFS, there are 11 secondary and 10 exploratory aims. The secondary aims focus on OS, PFS of subgroups, and event-free survival at 12 years, physician-reported treatment-related adverse event (AE) rates, patient-reported AE using pediatric and adult versions of Patient-Reported Outcomes version of the Common Terminology Criteria for Adverse Events (PRO-CTCAE), and the capture of representation by self-report and the use of electronic PRO data collections. The exploratory aims include the assessment of imaging central review as compared to local institutional review and the use of convolutional neural networks through artificial-intelligence machine learning. The study will examine the effect of therapy on acute and long-term adverse effects including health-related quality of life (HRQoL) and OS for up to 12 years. The study will compare patient self-report and provider AE reports and evaluate the associations of self-reported race/ethnicity and social determinants of health (SDOH)and post-progression/post-relapse OS. AHOD2131 strengthens the effort between the North American cooperative groups to conduct collaborative clinical trials and aims to harmonize an improved standard of care for ES cHL across the age continuum.

The Significance of a “MGUS” Tumor Board

Our large academic medical center consists of both a main campus and several regional sites that all offer hematology/oncology services where monoclonal gammopathy cases are referred for consultation. We receive well over 100 requests annually to see cases with abnormal monoclonal proteins. The challenge was how to triage these requests, with the goal of improving patient access to main campus myeloma experts while improving patient care and timely management. We decided to schedule cases initially referred for monoclonal gammopathy of undetermined significance (MGUS) with our advanced practice providers (APPs) both in the region and at main campus. Due to the concern for varying experience managing MGUS, differences in diagnostic practice and limited guidance we developed a novel concept to have a tumor board to review MGUS cases that were initially seen by our APPs. The “MGUS” tumor board is ran by three of our main campus myeloma focused hematologists and our myeloma dedicated APRN-PhD. We meet twice a month to review cases. Over the past 2 years we reviewed 147 cases providing recommendations for further work up. Of these, 41 cases (28%) were found to have a clinically significant monoclonal protein necessitating referral to a physician at main campus. The breakdown of these cases are detailed below in T able 1. Sixteen patients (11%) were started on treatment, the other 25 (17%) remained on observation. Another aspect of the tumor board is to provide education regarding clinically significant scenarios that involve monoclonal proteins. This tumor board offers CME credit for both APPs and physicians. By developing a team of knowledgeable APPs we have shifted 72% of monoclonal gammopathy referrals to remain under the care of APPs which improved access to main campus myeloma physician experts. This allowed the myeloma team at main campus to see more second opinions, open more clinical trials and expand our cell-therapy options for patients. Our referral base has grown and we have hired more APPs in the region and two additional myeloma physicians at main campus since starting the “MGUS” tumor board. Overall, we were successful in transitioning “MGUS” referrals to our APPs, but there were higher than expected percent of patients with clinically significant monoclonal proteins. Our data shows that it is essential for there to be some standardization for reviewing and working up “MGUS” referrals. This is especially important at large centers with multiple regional locations. We recommend a tumor board style review as it allows for discussion and education of “MGUS” referrals which prove to be far more complex in many cases.

The utility of word list and story recall for identifying older U.S. Chinese immigrants with cognitive impairment.

This study examined the utility of the Chinese-language translations of the word list memory test (Philadelphia Verbal Learning Test) and story memory test (Logical Memory subtest of the Wechsler Memory Scale) for differentiating cognitive diagnosis in older U.S. Chinese immigrants. Participants were ≥ 60 years old, with Chinese language proficiency to complete a diagnostic workup at the Mount Sinai's Alzheimer's Disease Research Center. The workup included an evaluation by a geriatric psychiatrist and cognitive testing with a psychometrician. Diagnosis of normal, mild cognitive impairment (MCI), and dementia was made independent of the cognitive tests at consensus led by a dementia expert physician. Multivariable logistic regression models were used to assess the sensitivity of story and word list memory tests for distinguishing between groups. Receiver operating characteristic (ROC area/area under the curve [AUC]) was used to compare the predictive accuracy of the two tests. The sample included 71 participants with normal cognition, 42 with MCI, and 24 with dementia. The MCI group was older and less educated than normal controls but younger and more educated than the dementia group. Delayed recall of both memory tests, but not immediate recall of either test, predicted diagnosis. While composite memory score of word list (AUC = 0.90) predicted diagnosis slightly better than that of stories (AUC = 0.85), the difference was not significant in this small sample (p = .14). Chinese-language translations of verbal memory tests, in particular delayed recall scores, were equally sensitive for classifying cognitive diagnosis in older U.S. Chinese immigrants. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Physician views of artificial intelligence in otolaryngology and rhinology: A mixed methods study.

The study aimed to investigate otolaryngologists' knowledge, trust, acceptance, and concerns with clinical applications of artificial intelligence (AI). This study used mixed methods with survey and semistructured interviews. Survey was e-mailed to American Rhinologic Society members, of which a volunteer sample of 86 members responded. Nineteen otolaryngologists were purposefully recruited and interviewed until thematic saturation was achieved. Seventy-six respondents (10% response rate) completed the majority of the survey: 49% worked in academic settings and 43% completed residency 10 or fewer years ago. Of 19 interviewees, 58% worked in academic settings, and 47% completed residency 10 or fewer years ago. Familiarity: Only 8% of survey respondents reported having AI training in residency, although 72% had familiarity with general AI concepts; 0 interviewees had personal experience with AI in clinical settings. Expected uses: Of the surveyed otolaryngologists, 82% would use an AI-based clinical decision aid and 74% were comfortable with AI proposing treatment recommendations. However, only 44% of participants would trust AI to identify malignancy and 53% to interpret radiographic images. Interviewees trusted AI for simple tasks, such as labeling septal deviation, more than complex ones, such as identifying tumors. Factors influencing AI adoption: 89% of survey participants would use AI if it improved patient satisfaction, 78% would be willing to use AI if experts and studies validated the technologies, and 73% would only use AI if it increased efficiency. Sixty-one percent of survey respondents expected AI incorporation into clinical practice within 5 years. Interviewees emphasized that AI adoption depends on its similarity to their clinical judgment and to expert opinion. Concerns included nuanced or complex cases, poor design or accuracy, and the personal nature of physician-patient relationships. Few physicians have experience with AI technologies but expect rapid adoption in the clinic, highlighting the urgent need for clinical education and research. Otolaryngologists are most receptive to AI "augmenting" physician expertise and administrative capacity, with respect for physician autonomy and maintaining relationships with patients. Level VI, descriptive or qualitative study.

SDA-Net: Self-distillation driven deformable attentive aggregation network for thyroid nodule identification in ultrasound images

Early detection and accurate identification of thyroid nodules are the major challenges in controlling and treating thyroid cancer that can be difficult even for expert physicians. Currently, many computer-aided diagnosis (CAD) systems have been developed to assist this clinical process. However, most of these systems are unable to well capture geometrically diverse thyroid nodule representations from ultrasound images with subtle and various characteristic differences, resulting in suboptimal diagnosis and lack of clinical interpretability, which may affect their credibility in the clinic. In this context, a novel end-to-end network equipped with a deformable attention network and a distillation-driven interaction aggregation module (DIAM) is developed for thyroid nodule identification. The deformable attention network learns to identify discriminative features of nodules under the guidance of the deformable attention module (DAM) and an online class activation mapping (CAM) mechanism and suggests the location of diagnostic features to provide interpretable predictions. DIAM is designed to take advantage of the complementarities of adjacent layers, thus enhancing the representation capabilities of aggregated features; driven by an efficient self-distillation mechanism, the identification process is complemented with more multi-scale semantic information to calibrate the diagnosis results. Experimental results on a large dataset with varying nodule appearances show that the proposed network can achieve competitive performance in nodule diagnosis and provide interpretability suitable for clinical needs.