Clinical Text Classification of Alzheimer’s Drugs’ Mechanism of Action

Kambar, Mina Esmail Zadeh Nojoo; Nahed, Pouyan; Cacho, Jorge Ramón Fonseca; Lee, Garam; Cummings, Jeffrey L.; Taghva, Kazem

doi:10.1007/978-981-16-2377-6_48

Cited by 15 publications

(3 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We used term frequency‐inverse document frequency (TF‐IDF) to generate the words’ matrix representation of the text and trained several ML models including XGBoost, random forest, logistic regression, decision tree, and support vector machine. The decision tree model achieved the highest accuracy (95%) in classifying the drugs’ MoA texts 17 . Human supervision verified the results and solved any unresolved identifications.…”

Section: Methodsmentioning

confidence: 59%

“…The decision tree model achieved the highest accuracy (95%) in classifying the drugs' MoA texts. 17 Human supervision verified the results and solved any unresolved identifications. In addition to ML models, we used multiple pattern extraction methods to extract information regarding biomarkers and other trials features from the trial's text description.…”

Section: Methodsmentioning

confidence: 76%

See 1 more Smart Citation

Alzheimer's disease drug development pipeline: 2022

Cummings

Lee

Nahed

et al. 2022

A&D Transl Res & Clin Interv

Self Cite

322

247

View full text Add to dashboard Cite

Introduction: Alzheimer's disease (AD) represents a global health crisis. Treatments are needed to prevent, delay the onset, slow the progression, improve cognition, and reduce behavioral disturbances of AD. We review the current clinical trials and drugs in development for the treatment of AD. Methods: We searched the governmental website clinicaltrials.gov where are all clinical trials conducted in the United States must be registered. We used artificial intelligence (AI) and machine learning (ML) approaches to ensure comprehensive detection and characterization of trials and drugs in development. We use the Common Alzheimer's Disease Research Ontology (CADRO) to classify drug targets and mechanisms of action of drugs in the pipeline. Results: As of January 25, 2022 (index date for this study) there were 143 agents in 172 clinical trials for AD. The pipeline included 31 agents in 47 trials in Phase 3, 82 agents in 94 trials in Phase 2, and 30 agents in 31 trials in Phase 1. Disease-modifying therapies represent 83.2% of the total number of agents in trials; symptomatic cognitive enhancing treatments represent 9.8% of agents in trials; and drugs for the treatment of neuropsychiatric symptoms comprise 6.9%. There is a diverse array of drug targets represented by agents in trials including nearly all CADRO categories. Thirtyseven percent of the candidate agents in the pipeline are repurposed drugs approved for other indications. A total of 50,575 participants are needed to fulfill recruitment requirements for all currently active clinical trials. Discussion: The AD drug development pipeline has agents representing a substantial array of treatment mechanisms and targets. Advances in drug design, outcome measures, use of biomarkers, and trial conduct promise to accelerate the delivery of new and better treatments for patients with AD.

show abstract

Section: Methodsmentioning

confidence: 59%

Section: Methodsmentioning

confidence: 76%

Alzheimer's disease drug development pipeline: 2022

Cummings

Lee

Nahed

et al. 2022

A&D Transl Res & Clin Interv

Self Cite

322

247

View full text Add to dashboard Cite

show abstract

“…The XGBoost model takes in the numerical vectors as input and outputs a predicted label based on the learned decision boundaries between the different classes, such as the presence or absence of any cancer findings within a pathology report. Such systems have been used for clinical report classification, such as identifying drug mechanisms 50 or identifying genetic mutations 51 .…”

Section: Rule-based Approachmentioning

confidence: 99%

Improving Information Extraction from Pathology Reports using Named Entity Recognition

Dutt

Witowski

et al. 2023

Preprint

View full text Add to dashboard Cite

Pathology reports are considered the gold standard in medical research due to their comprehensive and accurate diagnostic information. Natural language processing (NLP) techniques have been developed to automate information extraction from pathology reports. However, existing studies suffer from two significant limitations. First, they typically frame their tasks as report classification, which restricts the granularity of extracted information. Second, they often fail to generalize to unseen reports due to variations in language, negation, and human error. To overcome these challenges, we propose a BERT (bidirectional encoder representations from transformers) named entity recognition (NER) system to extract key diagnostic elements from pathology reports. We also introduce four data augmentation methods to improve the robustness of our model. Trained and evaluated on 1438 annotated breast pathology reports, acquired from a large medical center in the United States, our BERT model trained with data augmentation achieves an entity F1-score of 0.916 on an internal test set, surpassing the BERT baseline (0.843). We further assessed the model's generalizability using an external validation dataset from the United Arab Emirates, where our model maintained satisfactory performance (F1-score 0.860). Our findings demonstrate that our NER systems can effectively extract fine-grained information from widely diverse medical reports, offering the potential for large-scale information extraction in a wide range of medical and AI research. We publish our code at https://github.com/nyukat/pathology_extraction.

show abstract