Lung Cancer Prediction Model

Overview

Development stage

PilotSource: b) Pilot – The use case has been deployed in a limited test or pilot capacity.

High-impact designation

High-impactSource: a) High-impact

Impact justification

Not reported

Start date

Not reported

Withheld from public reporting

Not reported

Mission

Topic area

Health and Medical

Operational functions

Clinical and medical Derived by keyword rules; see methodology

Problem the AI is intended to solve

1. Treatment often proves difficult when diagnosed in advanced stages - long term outcomes are significantly improved when detected early. Screening programs are underutilized and definitively lacking in specific populations 2. This includes individuals at highest risk of lung cancer including US veterans. The Dayton VA Medical Center (DVAMC) initiative is intended to aid in detection of suspicious pulmonary nodules, expedite referrals for further evaluation and improve access to timely cancer care. In patients undergoing screening CTs, approx 50% of adults will have at least one lung nodule in their lifetime 3. Tools that provide aided detection and personalized risk of development of lung cancer may identify individuals that could require additional or more frequent screening. Sybil is a deep learning model to predict future lung cancer risk from a single low-dose chest computed tomography scan (LDCT) 4. The AI model utilizes one LDCT and the assistance of a radiologist to predict the risk of a patient developing lung cancer within six years. No additional clinical data is required. It has been validated on 3 independent data sets totaling almost 30K LDCTs. When a CT is processed by the algorithm, a series of 6 numbers corresponding to the patients aggregated risk of developing lung cancer in over 6 years. The hope is Sybil will allow lung cancer screening programs to be better utilized and promote increased care for at risk populations.

Expected benefits

When fully implemented, it can be used to personalize screening regime, calling high risk patients out earlier to identify cancers in the early stages and potentially reducing the screening burden on low risk patients.

System outputs

Descriptive statistics and basic analysis will be carried out using Excel. Descriptive statistics will analyze the data to be gathered as indicated, including ace, race, sex, and smoking status of patients. Sybil AI will calculate the sensitivity and specificity of predicted lung cancer rates to perform a receiver operating characteristic (ROC) analysis to validate the accuracy of the model. Power analysis may be performed to confirm the sample size needed with a significance level of 0.05 and a power of 0.80 using previously published area-under-the-curve data. Additional analysis such as logistic regression will be carried out using SPSS (Cary, NC). A statistician employed by Wright State University may assist with analysis of de-identified data only and after signing a dedicated data use agreement per the Dayton VAMC privacy policy.

Technology

AI classification

Classical / predictive MLSource: Classical/Predictive Machine Learning: Models trained on data to make predictions or classifications based on identified patterns or relatio…

System name(s)

Not reported

Custom-developed code

Yes

Public source code

Not reported

Sourcing

How it was built

Developed in-houseSource: b) Developed in-house

Vendor (as reported)

Not reported

Vendors (standardized)

None on the standard list

A vendor is the supplier named by the agency. It does not identify the underlying model or AI technology.

Data and privacy

Involves PII

Not reported

Privacy Impact Assessment

Not reported

Authorization to Operate

Not reported

Demographic features

Not reported

Training and evaluation data

Not reported

Federal Data Catalog

Not reported

Governance

0 of 8 minimum-practice questions answered. A blank answer means the agency reported nothing; it does not mean the practice is absent.

Potential impacts and how they were identified

Not reported