Assessing Perceptions of ML Explanations by Medical Oncologists

Survey of Medical Oncologists to Assess Trustworthiness of Various Approaches to AI Explainability for Prognostic Models

Status

Enrolling by invitation

Phases

Unknown

Study type

Interventional

Source

ClinicalTrials.gov

Registry ID

NCT06699615

Enrollment

Registered

2024-11-21

Start date

2024-12-02

Completion date

2028-04-15

Last updated

2026-04-06

For informational purposes only — not medical advice. Sourced from public registries and may not reflect the latest updates. Terms

Conditions

Oncology

Brief summary

The objective of this proposal is to conduct a vignette-based survey among practicing oncology clinicians who treat non-small cell lung cancer to assess the trustworthiness of explainable predictions from a neurosymbolic AI vs. State-of-the-art post-hoc explanatory algorithms, using simulated patient data.

Interventions

OTHEROnce time de-identified Qualtrics survey

Once time de-identified Qualtrics survey

Study design

Allocation

Intervention model

SINGLE_GROUP

Primary purpose

OTHER

Masking

NONE

Intervention model description

Consented participants will complete a one time de-identified Qualtrics survey.

Eligibility

Sex/Gender

ALL

Healthy volunteers

Inclusion criteria

"Medical oncologist" or "Hematologist/Oncologist" designation at Penn Treats lung cancer

Exclusion criteria

Non-medical oncologists No email address or physical address listed Does not treat lung cancer

Design outcomes

Primary

Measure	Time frame	Description
Determine whether neurosymbolic AI explainability methods improve the trustworthiness of explanations from a prognostic model, relative to post-hoc explainers.	3 months	Oncologist will be invited to complete a vignette-based survey

Countries

United States

Outcome results

None listed

Source: ClinicalTrials.gov · Data processed: Apr 7, 2026