Manual Review Research Articles

More Efficient Manual Review of Automatically Transcribed Tabular Data

Any machine learning method for transcribing historical text requires manual verification and correction, which is often time-consuming and expensive. Our aim is to make it more efficient. Previously, we developed a machine learning model to transcribe 2.3 million handwritten occupation codes from the Norwegian 1950 census. Here, we manually review the 90,000 codes (3%) for which our model had the lowest confidence scores. We allocated these codes to human reviewers, who used our custom annotation tool to review them. The reviewers agreed with the model's labels 31.9% of the time. They corrected 62.8% of the labels, and 5.1% of the images were uncertain or assigned invalid labels. 9,000 images were reviewed by multiple reviewers, resulting in an agreement of 86.4% and a disagreement of 9%. The results suggest that one reviewer per image is sufficient. We recommend that reviewers indicate any uncertainty about the label they assign to an image by adding a flag to their label. Our interviews show that the reviewers performed internal quality control and found our custom tool to be useful and easy to operate. We provide guidelines for efficient and accurate transcription of historical text by combining machine learning and manual review. We have open-sourced our custom annotation tool and made the reviewed images open access.

Getting to Know Named Entity Recognition: Better Information Retrieval

Named entity recognition (NER) is a powerful computer system that utilizes various computing strategies to extract information from raw text input, since the early 1990s. With rapid advancement in AI and computing, NER models have gained significant attention and been serving as foundational tools across numerus professional domains to organize unstructured data for research and practical applications. This is particularly evident in the medical and healthcare fields, where NER models are essential in efficiently extract critical information from complex documents that are challenging for manual review. Despite its successes, NER present limitations in fully comprehending natural language nuances. However, the development of more advanced and user-friendly models promises to improve work experiences of professional users significantly.

An explainable long short-term memory network for surgical site infection identification

BackgroundCurrently, surgical site infection surveillance relies on labor-intensive manual chart review. Recently suggested solutions involve machine learning to identify surgical site infections directly from the medical record. Deep learning is a form of machine learning that has historically performed better than traditional methods while being harder to interpret. We propose a deep learning model, a long short-term memory network, for the identification of surgical site infection from the medical record with an attention layer for explainability. MethodsWe retrieved structured data and clinical notes from the University of Utah Health System’s electronic health care record for operative events randomly selected for manual chart review from January 2016 to June 2021. Surgical site infection occurring within 30 days of surgery was determined according to the National Surgical Quality Improvement Program definition. We trained the long short-term memory model along with traditional machine learning models for comparison. We calculated several performance metrics from a holdout test set and performed additional analyses to understand the performance of the long short-term memory, including an explainability analysis. ResultsSurgical site infection was present in 4.7% of the total 9,185 operative events. The area under the receiver operating characteristic curve and sensitivity of the long short-term memory was higher (area under the receiver operating characteristic curve: 0.954, sensitivity: 0.920) compared to the top traditional model (area under the receiver operating characteristic curve: 0.937, sensitivity: 0.736). The top 5 features of the long short-term memory included 2 procedure codes and 3 laboratory values. ConclusionSurgical site infection surveillance is vital for the reduction of surgical site infection rates. Our explainable long short-term memory achieved a comparable area under the receiver operating characteristic curve and greater sensitivity when compared to traditional machine learning methods. With explainable deep learning, automated surgical site infection surveillance could replace burdensome manual chart review processes.

227 Spine-tuned Natural Language Models and Bespoke Regular Expression Classifiers for Automated Spinal Surgery Registry Development

INTRODUCTION: Surgical research demands the development of clinical registries, often through time-intensive manual chart review. Natural language processing (NLP) may accelerate registry development, and an ideal automatic registry (autoregistry) algorithm would be highly accurate while requiring minimal manual data annotation. NLP approaches including bespoke Regular Expression (RegEx) classifiers and Large Language Models (LLM) possess distinct strengths and weaknesses and have not been compared in the setting of autoregistry development. METHODS: We used an institutional data lake to retrieve 31,502 neurosurgical operative notes. A standardized set of spinal procedures was chosen for inclusion in the autoregistry. 200 manually annotated notes were used for training and testing purposes. RegEx classifiers were engineered to retrieve procedural info from unprocessed notes. A family of 110-million parameter BERT models, including LLM pre-trained on clinical text, was fine-tuned for the same tasks. We also tested a open-source 7-billion parameter LLM chatbot, Vicuna, without fine-tuning. RESULTS: The RegEx classifiers were able to identify spinal procedures and associated vertebral levels in nearly 99% of operative notes. Fine-tuned LLM identified common procedures (e.g. spinal fusion and laminectomy) with greater than 95% accuracy but performed poorly for rarer procedures (e.g. XLIF, corpectomy) and vertebral body identification. Qualitative evaluation of the Vicuna chatbot showed potential for the same tasks, following iteratively refined prompting. CONCLUSIONS: The goal of autoregistry development is to minimize time- and labor-intensive manual chart review. We found that fine-tuned LLM could not match the accuracy and efficiency of the RegEx classifier. However, LLM may be well-suited to expand existing clinical databases that provide a robust training set. Further work combining NLP approaches will attempt to develop a pipeline for autoregistry development from natural language (plain English) queries.

Diagnosis and Treatment of Perioperative Allergic Complications: A Practical Review.

Reported drug allergies are commonly encountered by surgeons and can lead to uncertainty in selecting an appropriate agent due to concerns of associated risks with related and cross-reactive drugs. This uncertainty can ultimately lead to increased infection rates. A literature review was conducted in PubMed using a combination of the terms "allergy," "allergic reaction," "anaphylaxis," and "surgery," "surgical," or "operating room" for articles published within the last 10 years. Publications identified with these search terms were then filtered for review articles, sorted by "best match," and a maximum of 100 articles were manually reviewed for each combination of search terms. Search results yielded 46,484 articles, 676 of which were ultimately included for manual review, based on selection criteria. Specifically, articles selected for inclusion focused on surgical allergic reactions that were either related to mechanism of action, causative agent for the allergic reaction, timing of allergic reaction, or recommendations for appropriate management. Allergic reactions can be a common occurrence in the operative room. Knowledge of likely causative agents, timing of a reaction to various agents, and appropriate management in the immediate and delayed setting can improve outcomes and safety for plastic surgery patients.

Predictive factors of basic palliative and hospice care among patients with cancer visiting the emergency department in a Hungarian tertiary care center

IntroductionPatients with advanced cancer tend to utilize the services of the health care system, particularly emergency departments (EDs), more often, however EDs aren't necessarily the most ideal environments for providing care to these patients. The objective of our study was to analyze the clinical and demographic characteristics of advanced patients with cancer receiving basic palliative care (BPC) or hospice care (HC), and to identify predictive factors of BPC and HC prior to their visit to the ED, in a large tertiary care center in Hungary. MethodsA retrospective, detailed analysis of patients receiving only BPC or HC, out of 1512 patients with cancer visiting the ED in 2018, was carried out. Sociodemographic and clinical data were collected via automated and manual chart review. Patients were followed up to determine length of survival. Descriptive and exploratory statistical analyses were performed. ResultsHospital admission, multiple (≥4x) ED visits, and respiratory cancer were independent risk factors for receiving only BPC (OR: 3.10, CI: 1.90–5.04; OR: 2.97, CI: 1.50–5.84; OR: 1.82, CI: 1.03–3.22, respectively), or HC (OR: 2.15, CI: 1.26–3.67; OR: 4.94, CI: 2.51–9.71; OR: 2.07, CI: 1.10–3.91). Visiting the ED only once was found to be a negative predictive factor for BPC (OR: 0.28, CI: 0.18–0.45) and HC (OR: 0.18, 0.10–0.31) among patients with cancer visiting the ED. ConclusionsOur study is the first from this European region to provide information regarding the characteristics of patients with cancer receiving BPC and HC who visited the ED, as well as to identify possible predictive factors of receiving BPC and HC. Our study may have relevant implications for health care planning strategies in practice.

Development of an Automatic Rule-Based Algorithm for the Detection of Ovarian Cancer Recurrence From Electronic Health Records.

As the onset of cancer recurrence is not explicitly recorded in the electronic health record (EHR), a high volume of manual chart review is required to detect the cancer recurrence. This study aims to develop an automatic rule-based algorithm for detecting ovarian cancer (OC) recurrence on the basis of minimally preprocessed EHR data. The automatic rule-based recurrence detection algorithm (Auto-Recur), using notes on image reading (positron emission tomography-computed tomography [PET-CT], CT, magnetic resonance imaging [MRI]), biomarker (CA125), and treatment information (surgery, chemotherapy, radiotherapy), was developed to detect the first OC recurrence. Auto-Recur contains three single algorithms (images, biomarkers, treatments) and hybrid algorithms (combinations of the single algorithms). The performance of Auto-Recur was assessed using sensitivity, specificity, and accuracy of the recurrence time detected. The recurrence-free survival probabilities were estimated and compared with the retrospective chart review results. The proposed Auto-Recur considerably reduced human resources and time; it saved approximately 1,340 days when scaled to 100,000 patients compared with the conventional retrospective chart review. The hybrid algorithm on the basis of a combination of image, biomarker, and treatment information was the most efficient (sensitivity: 93.4%, specificity: 97.4%) and precisely captured recurrence time (average time error: 8.5 days). The estimated 3-year recurrence-free survival probability (44%) was close to the estimates by the retrospective chart review (45%, log-rank P value = .894). Our rule-based algorithm effectively captured the first OC recurrence from large-scale EHR while closely approximating the recurrence-free survival estimates obtained by conventional retrospective chart reviews. The study findings facilitate large-scale EHR analysis, enhancing clinical research opportunities.

Assessment of Claimant, Clinical, and Financial Characteristics of Teleradiology Medical Malpractice Cases.

Background The increasing use of teleradiology has been accompanied by concerns relating to risk management and patient safety. Purpose To compare characteristics of teleradiology and nonteleradiology radiology malpractice cases and identify contributing factors underlying these cases. Materials and Methods In this retrospective analysis, a national database of medical malpractice cases was queried to identify cases involving telemedicine that closed between January 2010 and March 2022. Teleradiology malpractice cases were identified based on manual review of cases in which telemedicine was coded as one of the contributing factors. These cases were compared with nonteleradiology cases that closed during the same time period in which radiology had been determined to be the primary responsible clinical service. Claimant, clinical, and financial characteristics of the cases were recorded, and continuous or categorical data were compared using the Wilcoxon rank-sum test or Fisher exact test, respectively. Results This study included 135 teleradiology and 3474 radiology malpractices cases. The death of a patient occurred more frequently in teleradiology cases (48 of 135 [35.6%]) than in radiology cases (685 of 3474 [19.7%]; P < .001). Cerebrovascular disease was a more common final diagnosis in the teleradiology cases (13 of 135 [9.6%]) compared with the radiology cases (124 of 3474 [3.6%]; P = .002). Problems with communication among providers was a more frequent contributing factor in the teleradiology cases (35 of 135 [25.9%]) than in the radiology cases (439 of 3474 [12.6%]; P < .001). Teleradiology cases were more likely to close with indemnity payment (79 of 135 [58.5%]) than the radiology cases (1416 of 3474 [40.8%]; P < .001) and had a higher median indemnity payment than the radiology cases ($339 230 [IQR, $120 790-$731 615] vs $214 063 [IQR, $66 620-$585 424]; P = .01). Conclusion Compared with radiology cases, teleradiology cases had higher clinical and financial severity and were more likely to involve issues with communication. © RSNA, 2024 See also the editorial by Mezrich in this issue.

Brain Tumor Detection Using Classification Model

This study aims to develop automated methods for detecting and segmenting brain tumors in MRI scans using neural networks and conventional image processing. Accurately identifying tumor location and boundaries is crucial for diagnosis and determining the most effective treatment approach. Manual review by radiologists is time-consuming and subjective. However, recent advances in deep learning now enable more efficient analysis of large MRI datasets.&#x0D; A variety of neural network models have been explored for tumor segmentation, including recurrent networks, autoencoders and convolutional networks. RNNs are well-suited to capture contextual information across sequential MRI slices. Autoencoders perform unsupervised feature learning from the images. CNNs have shown promise due to their ability to directly learn visual patterns from pixel values. Unsupervised clustering techniques such as fuzzy c-means and k-means are also investigated to group tumor pixels based on intensity similarity without labeled data.&#x0D; Previous research combining these techniques has achieved high segmentation accuracy between 84-97% using additional image processing steps. Automated segmentation provides benefits over manual inspection such as reduced workload and more consistent delineation of tumor margins. It can precisely outline growths to aid radiologists in diagnosis and pathologists in assessing malignancy.&#x0D; The detection and segmentation of brain tumors has applications in both radiology and oncology. Radiologists could benefit from faster scan review and automated report generation. Neurosurgeons would gain insights into a tumor's size, location and infiltration to determine the most suitable surgical approach. Oncologists could utilize quantitative tumor features to select the most targeted radiation therapy or chemotherapy protocols. Overall, continued advances in deep and machine learning hold promise to improve diagnosis and management of brain cancer patients.

The hospital frailty risk score independently predicts postoperative outcomes in meningioma patients

ObjectiveThe Hospital Frailty Risk Score (HFRS) is a recently developed tool that uses ICD-10 codes to measure patient frailty. However, the effectiveness of HFRS has not yet been assessed in meningioma patients specifically. The present study aimed to evaluate the effectiveness of HFRS in predicting surgical outcomes for patients with meningiomas. MethodsThis retrospective study utilized data from patients undergoing meningioma resection at a single institution (2017–2019). Data were obtained through a combination of automated data retrieval and manual chart review. Bivariate logistic regression was used to assess the prognostic ability of several frailty indices for predicting postoperative outcomes. Further, discrimination for each model was assessed using the area under the receiver operating characteristic curve (AUROC). Generalized linear models with gamma error distributions and a log-link function were used to model hospital length of stay (LOS), total charges, complications, and disposition. ResultsA total of 464 meningioma patients (mean age 58.20 years, 72.8 % female, 66.4 % white) were included. HFRS had a significantly greater AUROC when compared to ASA (p = 0.0074) for postoperative complications, and HFRS significantly outperformed ASA (p = 0.0021) and mFI-5 (p = 0.018) when predicting nonroutine discharge. On multivariate analysis, increasing HFRS scores were significantly and independently associated with greater LOS (p < 0.0001), higher hospital charges (p < 0.0001), higher odds of postoperative complications (OR = 1.05, p = 0.019), and nonroutine discharge (OR = 1.12, p < 0.0001). The HFRS was non-inferior compared to the mFI-5, CCI, ASA and mFI-11 in terms of model discrimination. ConclusionHFRS effectively predicts postoperative outcomes for meningiomas and outperforms other indices in predicting complications and nonroutine discharge. This novel index may be used to improve clinical decision-making and reduce adverse postoperative outcomes among meningioma patients.

Overview of Molecular Diagnostics in Irish Clinical Oncology

Background Molecular diagnostics are critical for informing cancer patient care. In Ireland, the National Cancer Control Programme (NCCP) develops cancer therapy regimens, which include relevant information on molecular indications. Here, we present a collated overview of the current molecular indications of all NCCP systemic anti-cancer therapy regimens and the funding statuses of their associated drugs. Furthermore, we also provide estimates for the scale of required molecular testing in cancer therapy and for the clinical genetic sequencing capacity of Ireland, and provide a summary of current cancer clinical trials in Ireland which have molecular components. Methods Through a combination of web scraping, keyword search, and manual review, we performed a full review of all 757 indications included in the 476 therapy regimens published to date by the NCCP to identify therapy indications with explicit molecular criteria. For all cancer types identified in these indications, we obtained incidence rates in Ireland from National Cancer Registry Ireland to predict the number of patients yearly who stand to benefit from a molecular test. We then applied molecular subtype rates from published literature to estimate the number of patients who would then qualify for a relevant molecularly guided therapy. Results We identified 210 indications for 148 NCCP therapy regimens that include molecular criteria. These 210 molecular indications encompassed 85 genetic criteria, 137 cellular biomarker criteria, 57 molecularly informed drugs, and over 20 cancer types. We estimated that up to approximately 50% of cancer patients in Ireland could qualify for a molecular test and that the majority of tested patients would qualify for a treatment informed by a molecular test. Conclusions As personalised cancer medicine continues to develop in Ireland, this study will provide a baseline understanding of current practices. We anticipate that work such as this will help to inform planning in the healthcare system.

Machine learning natural language processing for identifying venous thromboembolism: systematic review and meta-analysis

AbstractVenous thromboembolism (VTE) is a leading cause of preventable in-hospital mortality. Monitoring VTE cases is limited by the challenges of manual medical record review and diagnosis code interpretation. Natural language processing (NLP) can automate the process. Rule-based NLP methods are effective but time consuming. Machine learning (ML)-NLP methods present a promising solution. We conducted a systematic review and meta-analysis of studies published before May 2023 that use ML-NLP to identify VTE diagnoses in the electronic health records. Four reviewers screened all manuscripts, excluding studies that only used a rule-based method. A meta-analysis evaluated the pooled performance of each study’s best performing model that evaluated for pulmonary embolism and/or deep vein thrombosis. Pooled sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) with confidence interval (CI) were calculated by DerSimonian and Laird method using a random-effects model. Study quality was assessed using an adapted TRIPOD (Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis) tool. Thirteen studies were included in the systematic review and 8 had data available for meta-analysis. Pooled sensitivity was 0.931 (95% CI, 0.881-0.962), specificity 0.984 (95% CI, 0.967-0.992), PPV 0.910 (95% CI, 0.865-0.941) and NPV 0.985 (95% CI, 0.977-0.990). All studies met at least 13 of the 21 NLP-modified TRIPOD items, demonstrating fair quality. The highest performing models used vectorization rather than bag-of-words and deep-learning techniques such as convolutional neural networks. There was significant heterogeneity in the studies, and only 4 validated their model on an external data set. Further standardization of ML studies can help progress this novel technology toward real-world implementation.

Extend the benchmarking indel set by manual review using the individual cell line sequencing data from the Sequencing Quality Control 2 (SEQC2) project

Accurate indel calling plays an important role in precision medicine. A benchmarking indel set is essential for thoroughly evaluating the indel calling performance of bioinformatics pipelines. A reference sample with a set of known-positive variants was developed in the FDA-led Sequencing Quality Control Phase 2 (SEQC2) project, but the known indels in the known-positive set were limited. This project sought to provide an enriched set of known indels that would be more translationally relevant by focusing on additional cancer related regions. A thorough manual review process completed by 42 reviewers, two advisors, and a judging panel of three researchers significantly enriched the known indel set by an additional 516 indels. The extended benchmarking indel set has a large range of variant allele frequencies (VAFs), with 87% of them having a VAF below 20% in reference Sample A. The reference Sample A and the indel set can be used for comprehensive benchmarking of indel calling across a wider range of VAF values in the lower range. Indel length was also variable, but the majority were under 10 base pairs (bps). Most of the indels were within coding regions, with the remainder in the gene regulatory regions. Although high confidence can be derived from the robust study design and meticulous human review, this extensive indel set has not undergone orthogonal validation. The extended benchmarking indel set, along with the indels in the previously published known-positive set, was the truth set used to benchmark indel calling pipelines in a community challenge hosted on the precisionFDA platform. This benchmarking indel set and reference samples can be utilized for a comprehensive evaluation of indel calling pipelines. Additionally, the insights and solutions obtained during the manual review process can aid in improving the performance of these pipelines.

Abstract 888: Uncovering clinically relevant omics signatures from pan-cancer imaging and multi-omics data integration

Abstract Despite advancements in deep learning for histopathology, integrating these insights with multi-omics data to uncover clinically relevant omics pathway-level signatures remains a challenge. Our study addresses this gap by applying unsupervised learning techniques on pan-cancer multi-omics data, leveraging 3,080 Hematoxylin and Eosin (H&amp;E) images from 1,010 patients in Clinical Proteomic Tumor Analysis Consortium (CPTAC) to uncover omics pathway-level signatures that drive discernable morphology phenotypes at the tissue level. First, imaging models were trained to predict clinical and mutation outcomes, and thereafter, integrated with transcriptomic and proteomic expression data using sparse canonical correlation analysis. Our findings reveal that images of TP53 mutated samples exhibited increased nuclear size and dense lymphoplasmacytic infiltration. These morphological changes correlated with neutrophil and macrophage signaling at the proteomic level, and IL-1 mediated signaling at the transcriptomic level, highlighting the complementary perspectives different omics can provide. To further elucidate immune interactions, we applied multi-omics deconvolution to identify 7 immune subtypes, characterizing each with gene set enrichment, germline DNA variations, and kinase activations. We show that imaging models trained to predict these subtypes differentiate tissue morphologies corresponding to immune enrichment (AUROC: 0.84). To further confirm the robustness of our approach, we trained models to predict co-regulated proteomic modules clustered by independent component analysis. We demonstrate a signature in pan-squamous tumors, consisting of T cell markers and interferon-gamma response proteins like CD4, CD48, and GBP5, which imaging models can successfully predict from histopathology. Manual review from pathologists confirmed lymphocytic T-cell density as a differentiator in samples with the highest and lowest T-cell signaling. Together, these approaches demonstrate that genomic characteristics discovered via unbiased mining of pan-cancer multi-omics data manifest as quantitative imaging phenotypes. These results underscore the potential of multi-omics and digital pathology to integratively uncover and confirm new cancer biology. Our ongoing work will explore predicting drug response and survival from morphology patterns related to multi-omics signatures. Citation Format: Joshua Wang, Runyu Hong, Jimin Tan, Wenke Liu, David Fenyȯ. Uncovering clinically relevant omics signatures from pan-cancer imaging and multi-omics data integration [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 888.

Mapping disability and climate change knowledge base in Scopus using bibliometric analysis.

Climate change and disability are rarely addressed by academic scholars within the spectrum of disabilities and as a single field of study. However, the intersectionality of disability exacerbates the vulnerability of people with disabilities to climate change as climate change frameworks in the Global North and South continue excluding them. This study aims to map the research-based knowledge housed in Scopus on disability and climate change. At the same time, it provides insights into innovative (novelty) ways of thinking and proposes a futuristic research agenda. A bibliometric analysis was conducted on Scopus-indexed articles using VOSviewer to map co-occurrences of keywords and co-authorship, and a manual thematic-scoping review augmented the data analysis. The disability and climate change debate as a joint study evolved from concern among health practitioners to human rights and social inclusion. In conclusion, there is a skewness towards mental health and medical sociology lens, while other sub-groups of persons with disabilities are yet to be engaged in co-creating disability-inclusive climate change knowledge. Thematic areas emerged as gaps that future studies embed principles enshrined in the United Nations Convention for the Rights of Persons with Disabilities and the Sustainable Development Goals.

Abstract 3569: Using AI to automatically process data from unstructured health records of patients with lung cancer

Abstract Introduction: Structured databases created from electronic health records (EHR) are crucial for cancer research. Manual data entry into databases is both labor-intensive and error-prone. We aimed to develop an artificial intelligence (AI)-driven approach for automatically inputting patient information from EHRs. Methods: A REDCap database for 53 patients with advanced lung cancer treated with first line immuno+/-chemotherapy at Gustave Roussy between 02/2021-06/2023 was manually populated by physicians with demographics, risk factors, cancer history, and treatment data, with 137 variables/patient. Given unstructured medical letters and a schematic description of each variable, generative AI was used to find, quote and process variables into a structured form. We directed large language model actions with prompt engineering and tailored few-shot examples. Mortality data were auto-extracted from the French public registry, INSEE. We assessed consistency between manual (MDE) and automated data entry (ADE), with a secondary manual review for mismatches performed by a senior physician. Results: In total, 7,261 data points were assessed. ADE averaged 10 minutes per patient inclusive of quality checks vs 1 h for MDA. The concordance rate between ADE and MDE was 90.2% (6,550/7,261) with a discordance of 6.8% (496/7,261). Data was missing in 0.8% (59/7,261) of cases for both methods, 1% (73/7,261) for ADE and 1,1% (83/7,261) for MDA. After checking discordances, ADE correctness was 95.3% (6,922/7,261) (Table), MDE was 93.8% (6,809/7,261). Errors in ADE (207/7,261) were due to algorithm refinement needs (70%) and missing EHR information (30%). Conclusion: Generative AI has a strong potential for identifying and structuring data from EHRs, yielding consistency superior to manual entry by physicians and 85% reduced amount of time. This may enhance the efficiency, accuracy, and scalability of EHR-to-database conversions. ADE for &amp;gt;1000 patients will be presented at the meeting. ADE correctness 100% 95-99% 90-94% 80-89% Sex; Age; Life status; Last follow-up; Metastatic from diagnosis Molecular alterations; Metastatic sites; Progression sites; Treatment type Tobacco &amp; cannabis consumption; Histology; PDL1 score; Progression event Pack years; Date of diagnosis &amp; first metastasis; Stage; Date of start treatment Citation Format: Mihaela Aldea, Pierre Rolland, Solenne Simon, Aliette Poplu, Muriel Wartelle, Benjamin Vignal, Jean-Charles Louis, Francois Lion, Arnaud Borie, David Planchard, Caroline Robert, Stefan Michiels, Fabrice Andre, Fabrice Barlesi, Franck Le Ouay, Benjamin Besse. Using AI to automatically process data from unstructured health records of patients with lung cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 3569.

Abstract 1771: NYU Langone Genome PACT - Genome Profiling of Actionable Cancer Targets (LG-PACT) for clinical patient molecular diagnostics and treatment

Abstract Next-generation sequencing (NGS) for the detection of somatic variants has become the tool of choice in a variety of molecular oncology fields and in the clinic. Its use ranges from sequencing entire tumor genomes and transcriptomes to targeted clinical diagnostic gene panels. The NYU Langone Genome PACT assay is a qualitative in vitro diagnostic test that uses targeted next generation sequencing of formalin-fixed paraffin-embedded (FFPE) tumor tissue matched with normal specimens from patients to detect gene alterations in a 607-gene panel. Indications for testing are cancer (solid tumors or hematological malignancies) where a mutational profile from multiple genes would be informative for disease stratification, prognosis, or treatment options including targeted therapies and eligibility for clinical trials. The test is intended to provide information on somatic mutations including point mutations, copy number aberrations, and small insertions/deletions (indels) for diagnostic and treatment decisions. LG-PACT is a United States Food and Drug Administration (FDA) approved diagnostic test. The clinical interpretation of sequencing data of molecular tumor markers from NGS encompasses automated variant calling tools with human interpretation. The final mostly manual review of data is intensive, involving highly trained scientists, encompassing literature review, interpretation and tier classification by pathologists, who then provide a complete molecular diagnostic report to the treating oncologists. Since January 2022 to present (October 2023) we have provided clinical genomic reports for 1029 oncological cases from 124 different cancers and their subtypes, including Brain (489 cases incl. meningioma, glioma and glioblastoma), Gastrointestinal (123), Lung (114), among others. We first present the technical challenges of validating an NGS oncological diagnostic targeted assay for appropriate clinical grade accuracy and sensitivity acceptable for patient care. We show how copy number alterations provide a more comprehensive description of the tumors genomic profile. We then outline the actionability of targeted panel sequencing for our current patient cohort. Where analysis of variant detection has led to 47% (490) of our clinical tumor samples containing known tier 1 therapeutic variants. We conclude with presenting case studies that identify both the clinical utility and informatic challenges of variant calling specific to gene panel sequencing e.g., i) the capture of potential targetable rare and novel indels and multivariant mutations in exons 19 and 20 of EGFR, and ii) a unique KIT tandem duplication event in a gastrointestinal stromal tumor (GIST). We demonstrate the value in precision oncology for multiple cancer types and how the capture of unique variants can provide better targeted treatment options to cancer patients. Citation Format: Varshini Vasudevaraja, Yiying Yang, Jonathan Serrano, Kazimierz Wrzeszczynski, Matija Snuderl. NYU Langone Genome PACT - Genome Profiling of Actionable Cancer Targets (LG-PACT) for clinical patient molecular diagnostics and treatment [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 1771.

Characteristics and outcomes of prehospital and emergency department surgical airways.

The surgical airway is a high acuity, low occurrence procedure. Data on the complications and outcomes of surgical airways are limited. Our primary objective was to describe immediate complications, late complications, and clinical outcomes of patients who underwent a surgical airway procedure in the prehospital or emergency department (ED) setting. We conducted a retrospective chart review of patients ≥14 years at an academic medical center who underwent a surgical airway procedure in the ED, the prehospital setting, or at a referring ED prior to interfacility transfer. We identified cases from keyword searches of prehospital text pages and hospital electronic medical records from June 1, 2008 to July 1, 2022. Manual chart review was used to confirm inclusion and determine patient and procedure characteristics. Outcomes included immediate complications, delayed in-hospital complications, and neurologic disability as defined by Modified Rankin Score (mRS) at discharge. We identified 63 patients (34 prehospital, 11 ED, and 18 referring ED). Immediate complications included mainstem intubation (46.0%) and bleeding that required direct pressure (23.4%). Overall, 29 patients (46%) died after arrival to the hospital. Of the patients surviving to hospital admission, 25 (48%) had an airway-related complication. Nine complications were deemed directly related to technical components of the procedure. Of the patients who survived to discharge, 18 (52.9%) had poor neurologic function (mRS 4-5). Procedural complications, mortality, and poor neurologic function were common following a surgical airway procedure in the prehospital or ED setting. Most patients surviving to discharge had a moderate to severe neurologic disability.

Using Electronic Health Record Data to Determine the Safety of Aqueous Humor Liquid Biopsies for Molecular Analyses

PurposeKnowing the surgical safety of anterior chamber liquid biopsies will support the increased use of proteomics and other molecular analyses to better understand disease mechanisms and therapeutic responses in patients and clinical trials. Manual review of operative notes from different surgeons and procedures in electronic health records (EHR) is cumbersome, but free-text software tools could facilitate efficient searches. DesignRetrospective case series. Participants1,418 aqueous humor (AH) liquid biopsies from patients undergoing intraocular surgery. MethodsFree-text EHR searches were performed using the Stanford Research Repository (STARR) cohort discovery tool to identify complications associated with anterior chamber paracentesis and subsequent endophthalmitis. Complications of the surgery unrelated to the biopsy were not reviewed. Main Outcome MeasuresBiopsy associated intraoperative complications and endophthalmitis. Results1,418 AH liquid biopsies were performed by 17 experienced surgeons. EHR free-text searches were 100% error-free for surgical complications, >99% for endophthalmitis (<1% false positive), and >93.6% for anesthesia type, requiring manual review for only a limited number of cases. More than 85% of cases were performed under local anesthesia without ocular muscle akinesia. Although the most common indication was cataract (50.1%), other diagnoses included glaucoma, diabetic retinopathy, uveitis, age-related macular degeneration, endophthalmitis, retinitis pigmentosa, and uveal melanoma. A 50-100μL sample was collected in all cases using either a 30-gauge needle or a blunt cannula via a paracentesis. The median follow-up was more than seven months. There was only one minor complication (0.07%) identified: a case of a small tear in Descemet’s membrane without long-term sequelae. No other complications occurred, including other corneal injuries, lens or iris trauma, hyphema, or suprachoroidal hemorrhage. There was no case of postoperative endophthalmitis. ConclusionsAnterior chamber liquid biopsy during intraocular surgery is a safe procedure and may be considered for large-scale collection of AH samples for molecular analyses. Free-text EHR searches are an efficient approach to reviewing intraoperative procedures.

The emergency department trigger tool: Multicenter trigger query validation.

We previously described derivation and validation of the emergency department trigger tool (EDTT) for adverse event (AE) detection. As the first step in our multicenter study of the tool, we validated our computerized screen for triggers against manual review, establishing our use of this automated process for selecting records to review for AEs. This is a retrospective observational study of visits to three urban, academic EDs over 18 months by patients ≥ 18 years old. We reviewed 912 records: 852 with at least one of 34 triggers found by the query and 60 records with none. Two first-level reviewers per site each manually screened for triggers. After completion, computerized query results were revealed, and reviewers could revise their findings. Second-level reviewers arbitrated discrepancies. We compare automated versus manual screening by positive and negative predictive values (PPVs, NPVs), present population trigger frequencies, proportions of records triggered, and how often manual ratings were changed to conform with the query. Trigger frequencies ranged from common (>25%) to rare (1/1000) were comparable at U.S. sites and slightly lower at the Canadian site. Proportions of triggered records ranged from 31% to 49.4%. Overall query PPV was 95.4%; NPV was 99.2%. PPVs for individual trigger queries exceeded 90% for 28-31 triggers/site and NPVs were >90% for all but three triggers at one site. Inter-rater reliability was excellent, with disagreement on manual screening results less than 5% of the time. Overall, reviewers amended their findings 1.5% of the time when discordant with query findings, more often when the query was positive than when negative (47% vs. 23%). The EDTT trigger query performed very well compared to manual review. With some expected variability, trigger frequencies were similar across sites and proportions of triggered records ranged 31%-49%. This demonstrates the feasibility and generalizability of implementing the EDTT query, providing a solid foundation for testing the triggers' utility in detecting AEs.