Can Artificial Intelligence Help Us See the Person Behind the Pattern? Intimate Partner Violence Through the Lens of Radiology - AJR Podcasts | Lyssna här

What do you do when images tell a different story? Bharti Khurana, MD, MBA, speaks with cohosts Lindsey Negrete, MD, and Amy Maduram, MD, about her journey from the reading room to the forefront of artificial intelligence innovation, transforming the way we identify and support survivors of intimate partner violence. Listen to their discussion in episode 4 of Extreme Radiology, an AJR Podcast Series.

https://www.ajronline.org/doi/10.2214/AJR.26.35966

*Key Takeaways

The Longitudinal Biomarker: There is no single characteristic injury that defines intimate partner violence (IPV). Instead, the true warning sign is a longitudinal pattern of trauma over time, including target injuries to the head and face, defensive injuries to the upper extremities, and neck injuries from strangulation.

Protecting the Patient Portal: Radiologists should not write "suspicious for IPV" in a formal imaging report because the partner may be monitoring the patient's electronic records. Radiologists should objectively describe the injuries in the text and directly call the referring clinician to raise abuse suspicions.

The "Pickle Jar" of Disclosure: If a case is flagged and the patient denies the abuse, the radiologist should not feel that they have failed. Reporting IPV is like opening a tight pickle jar; each safe conversation loosens the lid until the patient is finally ready and empowered to disclose their situation.

Humanizing Medicine with AI: The AIRS tool uses computational analysis to flag high probability IPV risk, which allows clinicians to better dedicate appointment times toward empathetic, trauma-informed conversations.

Bias as a Strength: Critics initially warned that radiologists could not diagnose IPV because they may not see the patient's socioeconomic background in the reading room. However, this is also a source of strength, allowing radiologists to report objective findings without certain potential implicit biases.

*Chapters

0:00 - Welcome and Content Warning

2:46 - Why Study IPV in Radiology

6:17 - Imaging Clues and Patterns

8:03 - Longitudinal Data and Integration

9:19 - Gender Differences and Misconceptions

14:33 - Reporting Safely in the EMR Era

19:00 - The Longitudinal Biomarker Eureka Moment

22:56 - AI That Prompts Human Care

25:21 - AIRS Workflow and Future Tools

30:25 - Collaboration and Leaving the Silo

34:47 - Handling Criticism and Staying Motivated