Child Abuse and Neglect | ISSUES | The Ohio State University

Detecting Hidden Mechanisms of Abuse-Related Injury in Children: Structured Topic Modeling of Clinical Death Reviews and Legal Opinions

Mon, 24 Aug 2026 00:00:00 +0000

Presented at the ISPCAN World Congress on Child Abuse and Neglect, Melbourne, Australia, August 2426, 2026.

Full title: Detecting Hidden Mechanisms of Abuse-Related Injury in Children: Structured Topic Modeling of Clinical Death Reviews and Legal Opinions Triangulating Multidisciplinary Child Fatality Reviews with Factual and Procedural Narratives from Case Law

Abstract

Existing child maltreatment surveillance systems miss or misclassify between 8090% of maltreatment deaths, and CPS records undercount fatalities by 5576%. Unstructured narrative text clinical death reviews, court opinions, and social work records accounts for 8090% of all organizational data produced, yet it is rarely leveraged for injury surveillance.

This presentation introduces a two-study AI + Structured Topic Modeling (STM) pipeline that integrates two distinct narrative corpora to recover latent, abuse-related injury mechanisms that structured administrative codes routinely obscure.

Study Design

	Study 1: CFRT Death Reviews	Study 2: CourtListener Case Law
Source	Colorado Child Fatality Review Team reports	Appellate opinions retrieved from CourtListener
Documents	223 child-level narrative records	751 opinions → 335 confirmed cases from Ohio
Text	Two free-text fields per report (clinical + administrative)	Facts + procedural history
LLM Extraction	GPT-4o-mini: caregiver behavior, injury mechanism, setting, diagnosis	GPT-4o-mini: 9 structured fields including perpetrator, victim type, legal outcome
STM Models	Risk & Contributing Factors; Circumstances	Injury Mechanism; Long-Term Consequences
Covariate	Incident year (temporal prevalence)	Judicial ruling (affirmed / reversed)

Research Questions

Can large language models transform unstructured child fatality narratives into standardized semantic features caregiver behavior, injury mechanism, setting, diagnosis?
What latent themes emerge, and what typology of risk and incident circumstances do they suggest?
Do the recovered themes reveal recurring patterns of head and neck trauma consistent with abuse-related injury?
How does the prevalence of these themes shift across incident years (20202025)?

Key Findings

Both corpora converge on craniofacial injury as the dominant mechanism routinely undercounted by ICD codes.
Strangulation emerged as a critical lethality predictor; soft-tissue damage can surface 23 days post-injury, requiring clinician and court training on delayed sequelae.
Records document long-term neurological harm (cerebral palsy, pituitary damage, blindness) in children often years before clinical follow-up.
Children represent a structurally distinct injury profile in both datasets, warranting specialized clinical and legal protocols.
“Accidental” injuries (falls, environmental contact) are frequently embedded in an abuse context screening for mechanism, not just diagnosis, is essential.
Appellate opinions and death reviews function as injury-surveillance sources beyond structured administrative fields.

Methodology

The pipeline applies:

PDF parsing (pdftools) and regex isolation of narrative fields
GPT-4o-mini for LLM-based structured field extraction
Structured Topic Modeling (STM) with incident year or judicial ruling as a prevalence covariate
BART zero-shot classification as an independent supervised-style validation check
FREX word scoring (frequency � exclusivity) for topic labeling

Lab Members

This presentation was developed by the ISSUES Lab team at The Ohio State University:

Dr. Gia E. Barboza-Salerno (Founding Director), Dr. Henriikka Weir, Malcolm McCarthy, Dr. Karla Shockley McCarthy, Dr. Amy Watson-Grace, Dr. Keith Warren, Esenia Cassidy, and Olivia McLucas with collaborators Scottye Cash and Suzanne Perkins.

📄 Presentation

Download the presentation slides here .

Examining the Interplay Between the Natural and Socio-physical Environment Using Machine Learning Algorithms

Thu, 01 Aug 2024 00:00:00 +0000

Plain-Language Summary

This research presentation examines the interplay between the built, natural, and social environments in predicting the risk of child abuse and neglect (CAN). By utilizing machine learning algorithms, the study categorizes environmental determinants of child well-being into key dimensions including urban design, environmental contaminants, and neighborhood socioeconomic conditions.

The study demonstrates that characteristics of the physical and natural environment, such as neighborhood walkability, lighting, and greenness, interact with social vulnerabilities to shape family stress and safety. This work highlights the value of using computational models to translate physical streetscape characteristics into policy-relevant indicators for child maltreatment prevention.

A Community-Based Approach to the Opioid Crisis

Mon, 01 Jan 0001 00:00:00 +0000

In progress

Real Help–Real Hope: Where Do We Go From Here?

A Community-Based Approach to the Opioid Crisis

Summary:
This keynote presentation, delivered at the Opioid Treatment Providers of Georgia conference, highlights the need for a comprehensive, public health–centered response to the opioid crisis. It emphasizes how social, environmental, and structural factors—such as poverty, racial inequities, and limited healthcare access—shape opioid-related harms, and critiques punitive policies in favor of community-driven, evidence-based solutions.

Key Details:


Short description:	This presentation critiques the criminalization-focused “War on Drugs” and outlines a public health framework for addressing the opioid crisis, emphasizing prevention, harm reduction, structural analysis, and equity-focused solutions for Georgia and beyond.
Lead developer:	Dr. Gia Elise Barboza-Salerno
Further reading:	Download the presentation slides for full details.
Main data source(s):	Opioid-related mortality, overdose surveillance, Naloxone distribution, treatment access data, regulatory frameworks, and structural vulnerability indicators for Georgia and the United States.
Coverage:	Georgia, with comparisons to national opioid trends.
Citation:	Barboza-Salerno, G.E. (2024). Real Help–Real Hope: Where Do We Go From Here? A Community-Based Approach to the Opioid Crisis. Presented at the Opioid Treatment Providers of Georgia Annual Conference, November 11, 2024.

Check out this accessibility and spatial equity tutorial for more details.

Examining the Interplay Between the Natural and Socio-physical Environment Using Machine Learning Algorithms

Mon, 01 Jan 0001 00:00:00 +0000

Our work on the socio-physical and natural enviornment was presented at the ISPCAN 2024 international conference.

No previous research has comprehensively examined how different aspects of the sociophysical and built environments increase child abuse and neglect (CAN) risk. Characteristics of the built and natural environment are underexplored in research on CAN. Five categories of environmental determinants of child well-being include: (i) urban design, (ii) contaminants, (iii) the parenting environment, (iv) socio-economic conditions, and (v) climate change1

Download the presentation here .

Integrating AI, Deep Learning, and Spatial Analytics to Investigate Environmental Correlates of Child Neglect

Mon, 01 Jan 0001 00:00:00 +0000

In progress

Integrating AI, Deep Learning, and Spatial Analytics to Investigate Environmental Correlates of Child Neglect

Summary:
The presentation discusses how to use computer vision to create indices of the built and perceptual environment and examine their association with child neglect.

Key Details:


Short description:	The presentation discusses how to use computer vision to create indices of the built and perceptual environment and examine their association with child neglect.
Lead developer:	Dr. Gia Elise Barboza-Salerno
Further reading:	Download the [presentation slides](/Lightning talk- nationwide.pdf) for full details.
Main data source(s):	ZenSVI, Mapillary, Area Deprivation Index, Tree Equity Scores.
Coverage:	Los Angeles, California.
Citation:	Barboza-Salerno, G.E. (2025). Integrating AI, Deep Learning, and Spatial Analytics to Investigate Environmental Correlates of Child Neglect. Presented at the 3rd Annual Family Violence Research-Practice Networking Event, Aug 13, 2025.

Check out the code using ZenSVI on our github page.

Utilizing AI-Driven Technology and Geospatial Analysis to Explore Landscape Morphology, Social Vulnerability, and Child Neglect in Los Angeles

Mon, 01 Jan 0001 00:00:00 +0000

Our work on Utilizing AI-Driven Technology and Geospatial Analysis to Explore Landscape Morphology, Social Vulnerability, and Child Neglect in Los Angeles will be presented at ISUF 2025 in Torino, Italy.

No previous research has comprehensively examined how different aspects of the sociophysical and built environments increase child abuse and neglect (CAN) risk. Characteristics of the built and natural environment are underexplored in research on CAN. Five categories of environmental determinants of child well-being include: (i) urban design, (ii) contaminants, (iii) the parenting environment, (iv) socio-economic conditions, and (v) climate change.

Child Abuse and Neglect | ISSUES | The Ohio State University

Detecting Hidden Mechanisms of Abuse-Related Injury in Children: Structured Topic Modeling of Clinical Death Reviews and Legal Opinions

Abstract

Study Design

Research Questions

Key Findings

Methodology

Lab Members

📄 Presentation

Examining the Interplay Between the Natural and Socio-physical Environment Using Machine Learning Algorithms

Plain-Language Summary

A Community-Based Approach to the Opioid Crisis

In progress

Real Help–Real Hope: Where Do We Go From Here?

Examining the Interplay Between the Natural and Socio-physical Environment Using Machine Learning Algorithms

Integrating AI, Deep Learning, and Spatial Analytics to Investigate Environmental Correlates of Child Neglect

In progress

A Novel Integration of Social, Perceptual, and Built Environmental Data

Utilizing AI-Driven Technology and Geospatial Analysis to Explore Landscape Morphology, Social Vulnerability, and Child Neglect in Los Angeles