Neoplasms

Respiratory Tract (Lung and Bronchial) Diseases

All Conditions

Lung Neoplasms

Machine learning algorithm run on EBUS images for real-time labelling of mediastinal lymph nodes and lymph node level

machine learning algorithm

Øyvind Ervik, MD

Hanne Sorger, MD,PhD

Håkon O Leira, MD, PhD

Øivind Rognmo, Dr.philos

To evaluate the usefulness of Deep neural network (DNN) in the evaluation of mediastinal and hilar lymph nodes with Endobronchial ultrasound (EBUS). The study will explore the feasibility of DNN to identify lymph nodes and blood vessel examined with EBUS.

Multi-center prospective feasibility study. The DNN model will be trained on ultrasound images with annotation to identifies lymph nodes and blood vessels examined with EBUS. The ability of the DNN to segment lymph nodes and vessels based on postoperative processing and static EBUS images will be evaluated in the first part of the study. In the second part of the study Real-time use of DNN in EBUS procedure will be evaluated.

Augmented Endobronchial Ultrasound (EBUS-TBNA) With Artificial Intelligence

Automatic Segmentation of Mediastinal Lymph Nodes and Blood Vessels in Endobronchial Ultrasound (EBUS) Images Using a Deep Neural Network

To explore if Deep neural network (DNN) has capability to segment lymph nodes and blood vessels from EBUS images

The precision the DNN has for detecting lymph nodes and blood vessels. Measured both per voxel in the EBUS images and per annotated structure (a structure is counted as detected if at least 50% of its annotated pixels are identified by the DNN).

True positive rate. Correctly detected lymph nodes/blood vessel over total lymph nodes/blood vessel. Measured per pixel in the EBUS images

Specificity = (True Negative)/(True Negative + False Positive). Measured per pixel in the EBUS images.

Measures the similarity between two sets of data: Annotated by pulmonologist vs DNN.

Is the run-time sufficiently low for real-time analysis during EBUS?

Procedure related adverse events or unexpected incidents registered

Helse Nord-Trøndelag HF

SINTEF Health Research

Norwegian University of Science and Technology

Outpatient participants evaluated as high risk by the machine learning algorithm and provided once weekly clinical evaluations

Outpatient participants evaluated as high risk by the machine learning algorithm and provided twice weekly clinical evaluations

machine learning directed identification of radiotherapy or chemoradiotherapy patients at high-risk for emergency department acute care and/or hospitalization

Machine learning algorithm

Manisha Palta, MD

This quality improvement project will evaluate the implementation of a previously described intervention (twice per week on-treatment clinical evaluations) in a feasible fashion using a previously described machine learning algorithm identifying patients identified at high risk for an emergency visit or hospitalization during radiation therapy.

Participants identified by the machine learning (ML) algorithm as high risk were randomized to either once weekly or twice weekly clinical evaluations

The ML directed twice-weekly evaluation arm was unblinded. Participants and providers were blinded to ML identification of high risk participants in the once weekly evaluation (standard of care) arm.

System for High-Intensity Evaluation During Radiotherapy

System for High Intensity EvaLuation During Radiation Therapy (SHIELD-RT): A Prospective Randomized Study of Machine Learning-directed Clinical Evaluations During Outpatient Cancer Radiation and Chemoradiation

De-identified summary data in the form of publication data tables and figures will be shared. Individual level data will not be shared.

Duke University

Rare Diseases

Acute Graft Versus Host Disease

machine learning directed identification of chemotherapy patients at high-risk for emergency department acute care and/or hospitalization

The objective of this study is to apply a validated machine-learning based model (SHIELD-RT, NCT04277650) to a cohort of patients undergoing systemic therapy as outpatient cancer treatment to generate an automatic system for the prediction of unplanned hospital admission rates and emergency department encounters.

A previously described machine learning (ML)-based model accurately predicted ED visits or hospitalizations for cancer patients undergoing radiation therapy or chemoradiation. An IRB approved prospective randomized trial, SHIELD-RT (NCT04277650) found that preemptive intervention for patients undergoing radiation and chemoradiation based on the ML model's risk stratification decreased the relative risk of acute care visits by 50%, showing that ML-guided escalation of care improved personalized supportive care and treatment compliance while decreasing healthcare costs.

The objective of this study is to apply this validated ML based model to a cohort of patients undergoing systemic therapy as outpatient cancer treatment to generate an automatic system for the prediction of unplanned hospital admission rates and emergency department encounters. Once validated, this study will add to the previously published body of evidence supporting a randomized trial evaluating the ML algorithm's ability to assign intervention for patients receiving systemic therapy at highest risk for acute care encounters.

Study title	NCT ID	Conditions	Interventions	Eligibility	Organization
Augmented Endobronchial Ultrasound (EBUS-TBNA) With Artificial Intelligence	NCT05739331	Artificial Inte... Endobronchial U... Lung Cancer	machine learnin...	18 Years -	Norwegian University of Science and Technology
System for High-Intensity Evaluation During Radiotherapy	NCT04277650	Radiation Thera... Chemotherapeuti...	Machine learnin...	18 Years -	Duke University
Machine Learning to Predict Acute Care During Cancer Therapy	NCT05122247	Chemotherapeuti...	Machine learnin...	18 Years -	Duke University
System for High-Intensity Evaluation During Radiotherapy	NCT04277650	Radiation Thera... Chemotherapeuti...	Machine learnin...	18 Years -	Duke University

Spots Global Cancer Trial Database for machine learning algorithm