Chatbot to Answer Radiology Based Procedure Patient Questions- Breast Imaging

Purpose	Facilitate patient education and patient - radiologist interaction by providing interactive patient education
Tag(s)	Non-Interpretative, Radiology Chatbot
Panel	Patient Facing
Define-AI ID	19110004
Originator	Patient Facing Panel
Lead	Michael Moore
Panel Chair	Michael Moore
Non-Interpretative Panel Chairs:	Alexander J Towbin, Adam Prate
Panel Reviewers	Patient Facing
License	Creative Commons 4.0
Status	Public Comment

                

Clinical Implementation

Value Proposition

A chatbot is a commonly employed technology designed to simulate human conversation utilizing either text based or auditory methods which has the potential to improve timely, patient-centric healthcare information dissemination. Currently, there is an absence of chatbot technology utilization specific to medical imaging. A radiology focused patient education chatbot is unique and with significant potential benefit to add value to our patients care. The primary value added would be providing accurate information to patients to supplement discussion with radiologists. This supplemental chatbot interaction could occur either prior to or after patient - radiologist interaction to facilitate understanding.

Narrative(s)

A 53- year-old woman presents to the breast imaging center for a yearly screening mammogram. A spiculated indistinct nodule is discovered in the left upper outer quadrant, determined to be a BI-RADs 4 lesion and biopsy is recommended.
Prior to a scheduled follow up appointment in 2 days for potential breast biopsy, the patient wished to ascertain accurate information about breast biopsy including how the procedure was performed and any potential complications. Utilizing a breast imaging focused patient chatbot, the patient gained an introductory understanding of the upcoming procedure which facilitated a robust in person informed consent conversation with the radiologist.

Workflow Description

The patient is provided with access to radiology-specific chatbot either at the time of scheduling, performing radiology examination or procedure, or as part of patient outreach efforts. Access could be provided directly at the radiology facility or by way of patient portal from health system, health system website, or social media. Access could be performed by several methods including smartphone or tablet devices.

Considerations for Dataset Development

Chatbot technology could be developed and employed for many specific areas within radiology, including varying subspecialties and exam - procedure modalities. For this use case focusing on breast imaging, examples include both stereotactic and ultrasound guided breast biopsies to evaluate for breast cancer. These occur in female patients (but not exclusively, as breast cancer in male patient occurs much more infrequently). This occurs most commonly in middle and older age demographics.

Dataset development would be best developed in conjunction with ACR member breast imagers to determine response to most frequent questions. Alternatively, utilizing patient center information with permission from reputable sources such as RadiologyInfo.org (ACR/RSNA sponsored).

Technical Specifications

Inputs

 

Questions from Patients, Text Based

Definition

Questions from patients potentially about any diagnostic tests or imaging-guided procedure in radiology.

Potential Value Set

Representative example questions relevant to this narrative could include: What are the most likely complications of an ultrasound-guided breast biopsy? How long is the recovery time from a stereotactic breast biopsy? What will happen during the biopsy procedure? Will I experience pain during or after the procedure? How long does the typical biopsy take? What resources are available to handle the stress of this situation? Are the accessible communities of patients/families who offer each other support for breast cancer? Will a family member be allowed with me during the procedure? How long will it take to obtain the results from this procedure? How do I access the results? Who will discuss the findings with me?

Primary Outputs

Text Based Answers to Patient Questions

Definition	Variety of text-based answers depending on underlying questions and/or imaging procedure
Data Type	Text
Potential Value Set	Representative example responses could be obtained from: Radiologists with expertise in a specific area of interest, in this example ACR member breast imagers. Patients and patient groups with particular knowledge and interest in the underlying disease and associated imaging. Publicly available sources (with permission) such as RadiologyInfo.org A fail safe output if the chatbot is stuck which could either redirect to beginning of chatbot menu or to a human provider.

Future Development Ideas

This is an open-ended use case scenario which could encompass any patient-centered interactive chatbot education for a radiology based procedure or diagnostic test. Potential steps could also include video-based content of educational material linked to underlying chatbot. An additional avenue may include the chatbot asking patients questions to facilitate interaction. For very complex questions, development of the option to speak with a human directly could be entertained. However, as the chatbot is for supplemental educational information to enhance (rather than replace) the radiologist-patient discussion this is beyond the scope of this use case. Future development beyond text-based chatbot into auditory based device could serve as an aspirational goal.

Related Datasets

No known related public datasets at this time,  please alert us if you know of any.