class: center, middle, inverse, title-slide .title[ #
Predictive Modeling of Racial Disparities in U.S. Violent Deaths
] .author[ ### Ying-Ju Tessa Chen, PhD
Scholar
|
@ying-ju
|
ychen4@udayton.edu
Joint work with:
Tatjana Miljkovic, PhD
Miami University
] .date[ ### August 3, 2025 | 2025 JSM | Nashville, TN ] --- ## Motivation & Research Questions <div style="margin-top:-1.2em;"></div> .pull-left[ .red[Surveillance of national violent deaths (Forsberg, Sheats, Blair, Nguyen, E., Betz, and Lyons (2025)):] + 2022: 74,148 violent deaths + Composition: - Suicides: 60.6% (n = 44,917) - Homicides: 30.2% (n = 22,395) - Deaths of undetermined intent 7.1% (n = 5,292) - Legal intervention deaths: 1.4% (n = 1,014) - Unintentional firearm deaths: < 1.0% (n = 530) ] .pull-right[ .red[Identified gaps:] - Heavy reliance on .blue[descriptive stats] and basic models - .blue[SDOH factors] often excluded from analysis - Limited focus on .blue[racial disparities] as central variables - .blue[Geographic (state-level) variation] underexplored - Lack of .blue[robustness checks] and model validation - NVDRS data .blue[underutilized or inconsistently applied] - Few studies follow a .blue[structured modeling framework] (e.g., DSF) - Minimal attention to .blue[business or policy applications] 🔍 .red[.small[These gaps motivate the following research questions →]] ] <div style="position: absolute; bottom: 10px; right: 20px; font-size: 0.8em; color: gray;"> 1 / 5 </div> --- ## Research Questions & Study Scope <div style="margin-top:-1.2em;"></div> .left-code[ 1. **Demographics** How do racial and age groups differ in violent death risk? 2. **State-Level Variation** Which states show higher-than-expected deaths? 3. **Mental Health** How does treatment access affect outcomes? 4. **Substance Use** What role does alcohol or drug use play? 5. **Environment** What contextual factors shape risk? ] .right-plot[ **NVDRS State Data - 46 States** <style> .custom h2, .custom p { margin-bottom: 0; } </style>
] <div style="position: absolute; bottom: 10px; right: 20px; font-size: 0.8em; color: gray;"> 2 / 5 </div> --- ## Modeling Approach <div style="margin-top:-1.2em;"></div> <img src="data:image/png;base64,#./figures/JBA_overview.png" width="90%" style="display: block; margin: auto;" /> <div style="position: absolute; bottom: 10px; right: 20px; font-size: 0.8em; color: gray;"> 3 / 5 </div> --- ## Summary of Model Findings <div style="margin-top:-1.2em;"></div> .pull-left-2[ .small[**Model-Specific Findings**] .footnotesize[ **Suicide** - Higher excess risk in NY, NJ, MA - A/PI group: 43% lower risk - Mental health problem → protective - Receiving treatment → ↑ risk - Death at home: significant **Homicide** - Males at higher risk - State-level differences: NY, NJ, MA again highest - A/PI group: lower risk - Housing instability (p < 0.1): potential risk factor **Other Manner of Death** - Older age groups: higher risk - Legal intervention deaths vs. undetermined: key contrast - NY, MN, OH → lower excess deaths ] ] .pull-right-2[ .small[**Cross-Cutting Patterns**] - Sex and Age only matter for homicide and other, not suicide - A/PI consistently show lower risk across categories - Mental health factors show complex patterns (protective vs. severity signals) - Clear state-level disparities → need region-specific interventions - Substance use not significant across models ] <div style="position: absolute; bottom: 10px; right: 20px; font-size: 0.8em; color: gray;"> 4 / 5 </div> --- ## Conclusion & Implication .pull-left-2[ **Takeaways** - Violent death patterns vary significantly across state, race, and manner of death - Mental health indicators are critical—but show complex effects (protective vs. severity) - Asian/Pacific Islanders show consistently lower risk across models - Substance use was not a significant predictor in final models - Demonstrated value of SDOH + Design Science Framework in structuring analysis ] .pull-right-2[ **Implications & Next Steps** - State-specific disparities call for region-targeted interventions - LR models offer interpretability, while ML adds predictive performance depth .blue[Limitations:] - Missing data from CA, TX, FL reduces generalizability - Sparse state data and lack of county-level detail limit granularity .blue[Future work:] expand datasets, incorporate local-level variables, and refine predictive pipelines ] <div style="position: absolute; bottom: 10px; right: 20px; font-size: 0.8em; color: gray;"> 5 / 5 </div> --- .pull-left[ <div style="margin-top:-1.2em;"></div> ### Reference <div style="margin-top:-1em;"></div> Chen, Y. J., & Miljkovic, T. (2025). Data-driven insights into racial disparities in violent deaths in the United States: predictive models for risk assessment and business solutions. *Journal of Business Analytics*, 1-28. ### Thanks <div style="margin-top:-1em;"></div> - This research is sponsored by the Diversity, Equity, and Inclusion Strategic Research Program in the Society of Actuaries Research Institute. - Please do not hesitate to contact me (Tessa Chen) at <a href="mailto:ychen@udayton.edu"><i class="fa fa-paper-plane fa-fw"></i> ychen4@udayton.edu</a> for questions or further discussions. ### Want to know more details? <div style="margin-top:-1em;"></div> Stop by the Music City Center in Hall B at the .red[poster number 10]. ] .pull-right[ <br> <br> <img src="data:image/png;base64,#./figures/Tessa_grey_G.gif" width="60%" style="display: block; margin: auto;" /> ] --- class: center, middle, inverse, title-slide .title[ # <p>Predictive Modeling of Racial Disparities in U.S. Violent Deaths</p> ] .author[ ### Ying-Ju Tessa Chen, PhD <br>[<svg viewBox="0 0 488 512" style="height:1em;position:relative;display:inline-block;top:.1em;fill:white;" xmlns="http://www.w3.org/2000/svg"> <path d="M488 261.8C488 403.3 391.1 504 248 504 110.8 504 0 393.2 0 256S110.8 8 248 8c66.8 0 123 24.5 166.3 64.9l-67.5 64.9C258.5 52.6 94.3 116.6 94.3 256c0 86.5 69.1 156.6 153.7 156.6 98.2 0 135-70.4 140.8-106.9H248v-85.3h236.1c2.3 12.7 3.9 24.9 3.9 41.4z"></path></svg> Scholar](https://scholar.google.com/citations?user=nfXnYKcAAAAJ&hl=en&oi=ao) | [<svg viewBox="0 0 496 512" style="height:1em;position:relative;display:inline-block;top:.1em;fill:white;" xmlns="http://www.w3.org/2000/svg"> <path d="M165.9 397.4c0 2-2.3 3.6-5.2 3.6-3.3.3-5.6-1.3-5.6-3.6 0-2 2.3-3.6 5.2-3.6 3-.3 5.6 1.3 5.6 3.6zm-31.1-4.5c-.7 2 1.3 4.3 4.3 4.9 2.6 1 5.6 0 6.2-2s-1.3-4.3-4.3-5.2c-2.6-.7-5.5.3-6.2 2.3zm44.2-1.7c-2.9.7-4.9 2.6-4.6 4.9.3 2 2.9 3.3 5.9 2.6 2.9-.7 4.9-2.6 4.6-4.6-.3-1.9-3-3.2-5.9-2.9zM244.8 8C106.1 8 0 113.3 0 252c0 110.9 69.8 205.8 169.5 239.2 12.8 2.3 17.3-5.6 17.3-12.1 0-6.2-.3-40.4-.3-61.4 0 0-70 15-84.7-29.8 0 0-11.4-29.1-27.8-36.6 0 0-22.9-15.7 1.6-15.4 0 0 24.9 2 38.6 25.8 21.9 38.6 58.6 27.5 72.9 20.9 2.3-16 8.8-27.1 16-33.7-55.9-6.2-112.3-14.3-112.3-110.5 0-27.5 7.6-41.3 23.6-58.9-2.6-6.5-11.1-33.3 2.6-67.9 20.9-6.5 69 27 69 27 20-5.6 41.5-8.5 62.8-8.5s42.8 2.9 62.8 8.5c0 0 48.1-33.6 69-27 13.7 34.7 5.2 61.4 2.6 67.9 16 17.7 25.8 31.5 25.8 58.9 0 96.5-58.9 104.2-114.8 110.5 9.2 7.9 17 22.9 17 46.4 0 33.7-.3 75.4-.3 83.6 0 6.5 4.6 14.4 17.3 12.1C428.2 457.8 496 362.9 496 252 496 113.3 383.5 8 244.8 8zM97.2 352.9c-1.3 1-1 3.3.7 5.2 1.6 1.6 3.9 2.3 5.2 1 1.3-1 1-3.3-.7-5.2-1.6-1.6-3.9-2.3-5.2-1zm-10.8-8.1c-.7 1.3.3 2.9 2.3 3.9 1.6 1 3.6.7 4.3-.7.7-1.3-.3-2.9-2.3-3.9-2-.6-3.6-.3-4.3.7zm32.4 35.6c-1.6 1.3-1 4.3 1.3 6.2 2.3 2.3 5.2 2.6 6.5 1 1.3-1.3.7-4.3-1.3-6.2-2.2-2.3-5.2-2.6-6.5-1zm-11.4-14.7c-1.6 1-1.6 3.6 0 5.9 1.6 2.3 4.3 3.3 5.6 2.3 1.6-1.3 1.6-3.9 0-6.2-1.4-2.3-4-3.3-5.6-2z"></path></svg> @ying-ju](https://github.com/ying-ju) | [<svg viewBox="0 0 512 512" style="height:1em;position:relative;display:inline-block;top:.1em;fill:white;" xmlns="http://www.w3.org/2000/svg"> <path d="M476 3.2L12.5 270.6c-18.1 10.4-15.8 35.6 2.2 43.2L121 358.4l287.3-253.2c5.5-4.9 13.3 2.6 8.6 8.3L176 407v80.5c0 23.6 28.5 32.9 42.5 15.8L282 426l124.6 52.2c14.2 6 30.4-2.9 33-18.2l72-432C515 7.8 493.3-6.8 476 3.2z"></path></svg> ychen4@udayton.edu](mailto:ychen4@udayton.edu)</br><br><u><b><font color="white">Joint work with:</b></u><br>Tatjana Miljkovic, PhD [<svg viewBox="0 0 512 512" style="height:1em;position:relative;display:inline-block;top:.1em;fill:white;" xmlns="http://www.w3.org/2000/svg"> <path d="M326.612 185.391c59.747 59.809 58.927 155.698.36 214.59-.11.12-.24.25-.36.37l-67.2 67.2c-59.27 59.27-155.699 59.262-214.96 0-59.27-59.26-59.27-155.7 0-214.96l37.106-37.106c9.84-9.84 26.786-3.3 27.294 10.606.648 17.722 3.826 35.527 9.69 52.721 1.986 5.822.567 12.262-3.783 16.612l-13.087 13.087c-28.026 28.026-28.905 73.66-1.155 101.96 28.024 28.579 74.086 28.749 102.325.51l67.2-67.19c28.191-28.191 28.073-73.757 0-101.83-3.701-3.694-7.429-6.564-10.341-8.569a16.037 16.037 0 0 1-6.947-12.606c-.396-10.567 3.348-21.456 11.698-29.806l21.054-21.055c5.521-5.521 14.182-6.199 20.584-1.731a152.482 152.482 0 0 1 20.522 17.197zM467.547 44.449c-59.261-59.262-155.69-59.27-214.96 0l-67.2 67.2c-.12.12-.25.25-.36.37-58.566 58.892-59.387 154.781.36 214.59a152.454 152.454 0 0 0 20.521 17.196c6.402 4.468 15.064 3.789 20.584-1.731l21.054-21.055c8.35-8.35 12.094-19.239 11.698-29.806a16.037 16.037 0 0 0-6.947-12.606c-2.912-2.005-6.64-4.875-10.341-8.569-28.073-28.073-28.191-73.639 0-101.83l67.2-67.19c28.239-28.239 74.3-28.069 102.325.51 27.75 28.3 26.872 73.934-1.155 101.96l-13.087 13.087c-4.35 4.35-5.769 10.79-3.783 16.612 5.864 17.194 9.042 34.999 9.69 52.721.509 13.906 17.454 20.446 27.294 10.606l37.106-37.106c59.271-59.259 59.271-155.699.001-214.959z"></path></svg> Miami University](https://miamioh.edu/profiles/cas/tatjana-milijkovic.html)<br><br/> ] .date[ ### August 3, 2025 | 2025 JSM | Nashville, TN ] --- name: appendix-start class: center, middle, inverse # Appendix --- ## Response Variable: Excess Mortality Indicator To compare mortality across demographic groups and states, we created a binary response variable based on the **Standardized Mortality Ratio (SMR)**. The SMR measures how observed deaths compare to expected deaths after adjusting for race, age, and sex: `$$\text{SMR}_{itras} = \frac{d_{itras}}{\tilde{d}_{itras}}, \quad \text{where } \tilde{d}_{itras} = \frac{D_{tras}}{P_{tras}} \times p_{itras}$$` - `\(d_{itras}\)`: observed deaths in state *i*, year *t*, race *r*, age group *a*, sex *s* - `\(\tilde{d}_{itras}\)`: expected deaths using national standard population rates We define the binary response variable as: `$$\mathbf{1}_{[\text{SMR}]} = \begin{cases} 1, & \text{if } \text{SMR} > 1 \quad \text{(excess deaths)} \\ 0, & \text{otherwise} \end{cases}$$` This indicator allows us to assess **risk of excess mortality** across demographic and geographic dimensions using logistic regression. <div style="position: absolute; bottom: 10px; right: 20px; font-size: 0.8em; color: gray;"> A - 1 </div> --- ## Sample Size & Number of Predictors <table class="table table-striped table-hover table-condensed" style="color: black; width: auto !important; margin-left: auto; margin-right: auto;"> <thead> <tr> <th style="text-align:left;"> Case </th> <th style="text-align:center;"> Suicide </th> <th style="text-align:center;"> Homicide </th> <th style="text-align:center;"> Other.Manners </th> </tr> </thead> <tbody> <tr> <td style="text-align:left;"> training data (2020) </td> <td style="text-align:center;"> 901 </td> <td style="text-align:center;"> 732 </td> <td style="text-align:center;"> 839 </td> </tr> <tr> <td style="text-align:left;"> test data (2021) </td> <td style="text-align:center;"> 912 </td> <td style="text-align:center;"> 739 </td> <td style="text-align:center;"> 881 </td> </tr> <tr> <td style="text-align:left;"> number of predictors </td> <td style="text-align:center;"> 19 </td> <td style="text-align:center;"> 17 </td> <td style="text-align:center;"> 21 </td> </tr> </tbody> </table> <div style="position: absolute; bottom: 10px; right: 20px; font-size: 0.8em; color: gray;"> A - 2 </div> --- ## Model Comparison - Suicide Model <table class="table table-striped table-hover table-condensed" style="color: black; width: auto !important; margin-left: auto; margin-right: auto;border-bottom: 0;"> <caption>Comparison of predictive models for suicide death</caption> <thead> <tr> <th style="text-align:left;"> Model </th> <th style="text-align:center;"> AUC </th> <th style="text-align:center;"> Accuracy </th> <th style="text-align:center;"> Sensitivity </th> <th style="text-align:center;"> Specificity </th> </tr> </thead> <tbody> <tr> <td style="text-align:left;"> LR </td> <td style="text-align:center;"> 0.7404 (0.0181) </td> <td style="text-align:center;"> 0.7314 (0.0147) </td> <td style="text-align:center;"> 0.8172 (0.0128) </td> <td style="text-align:center;"> 0.5406 (0.0165) </td> </tr> <tr> <td style="text-align:left;"> RF </td> <td style="text-align:center;"> 0.7749 (0.0170) </td> <td style="text-align:center;"> 0.7544 (0.0143) </td> <td style="text-align:center;"> 0.9316 (0.0084) </td> <td style="text-align:center;"> 0.3604 (0.0159) </td> </tr> <tr> <td style="text-align:left;"> GB </td> <td style="text-align:center;"> 0.7676 (0.0172) </td> <td style="text-align:center;"> 0.7522 (0.0143) </td> <td style="text-align:center;"> 0.8649 (0.0113) </td> <td style="text-align:center;"> 0.5018 (0.0166) </td> </tr> <tr> <td style="text-align:left;"> SVM </td> <td style="text-align:center;"> 0.7833 (0.0167) </td> <td style="text-align:center;"> 0.7489 (0.0144) </td> <td style="text-align:center;"> 0.8331 (0.0123) </td> <td style="text-align:center;"> 0.5618 (0.0164) </td> </tr> <tr> <td style="text-align:left;"> NN </td> <td style="text-align:center;"> 0.7802 (0.0170) </td> <td style="text-align:center;"> 0.7423 (0.0145) </td> <td style="text-align:center;"> 0.8045 (0.0131) </td> <td style="text-align:center;"> 0.6042 (0.0162) </td> </tr> </tbody> <tfoot><tr><td style="padding: 0; " colspan="100%"> <sup></sup> Note: Standard errors are included in parentheses.</td></tr></tfoot> </table> <br> **Final LR model includes these variables:** .small[ - State - Race - the proportion of deaths occurring at home - the proportion of individuals identified as currently having a mental health problem - the proportion of individuals currently receiving treatment for a mental health problem or substance abuse problem. ] <div style="position: absolute; bottom: 10px; right: 20px; font-size: 0.8em; color: gray;"> A - 3 </div> --- ## Model Comparison - Homicide Model <table class="table table-striped table-hover table-condensed" style="color: black; width: auto !important; margin-left: auto; margin-right: auto;border-bottom: 0;"> <caption>Comparison of predictive models for homicide death</caption> <thead> <tr> <th style="text-align:left;"> Model </th> <th style="text-align:left;"> AUC </th> <th style="text-align:left;"> Accuracy </th> <th style="text-align:left;"> Sensitivity </th> <th style="text-align:left;"> Specificity </th> </tr> </thead> <tbody> <tr> <td style="text-align:left;"> LR </td> <td style="text-align:left;"> 0.7933 (0.0168) </td> <td style="text-align:left;"> 0.7307 (0.0163) </td> <td style="text-align:left;"> 0.7758 (0.0153) </td> <td style="text-align:left;"> 0.6621 (0.0174) </td> </tr> <tr> <td style="text-align:left;"> RF </td> <td style="text-align:left;"> 0.8237 (0.0156) </td> <td style="text-align:left;"> 0.7537 (0.0158) </td> <td style="text-align:left;"> 0.8991 (0.0111) </td> <td style="text-align:left;"> 0.5324 (0.0184) </td> </tr> <tr> <td style="text-align:left;"> GB </td> <td style="text-align:left;"> 0.8250 (0.0151) </td> <td style="text-align:left;"> 0.7442 (0.0160) </td> <td style="text-align:left;"> 0.8117 (0.0144) </td> <td style="text-align:left;"> 0.6416 (0.0176) </td> </tr> <tr> <td style="text-align:left;"> SVM </td> <td style="text-align:left;"> 0.8146 (0.0166) </td> <td style="text-align:left;"> 0.7767 (0.0153) </td> <td style="text-align:left;"> 0.8094 (0.0144) </td> <td style="text-align:left;"> 0.7270 (0.0164) </td> </tr> <tr> <td style="text-align:left;"> NN </td> <td style="text-align:left;"> 0.8421 (0.0149) </td> <td style="text-align:left;"> 0.7876 (0.0150) </td> <td style="text-align:left;"> 0.8274 (0.0139) </td> <td style="text-align:left;"> 0.7270 (0.0164) </td> </tr> </tbody> <tfoot><tr><td style="padding: 0; " colspan="100%"> <sup></sup> Note: Standard errors are included in parentheses.</td></tr></tfoot> </table> <br> **Final LR model includes these variables:** .small[ - State - Sex - Age Group - the proportion of individuals experiencing acute or chronic instability in their housing situation, which appears to have contributed to their death - the proportion of victims precipitated by another serious crime, such as drug dealing or robbery ] <div style="position: absolute; bottom: 10px; right: 20px; font-size: 0.8em; color: gray;"> A - 4 </div> --- ## Model Comparison - Other Manner of Death Model <table class="table table-striped table-hover table-condensed" style="color: black; width: auto !important; margin-left: auto; margin-right: auto;border-bottom: 0;"> <caption>Comparison of predictive models for other manners of death</caption> <thead> <tr> <th style="text-align:left;"> Model </th> <th style="text-align:left;"> AUC </th> <th style="text-align:left;"> Accuracy </th> <th style="text-align:left;"> Sensitivity </th> <th style="text-align:left;"> Specificity </th> </tr> </thead> <tbody> <tr> <td style="text-align:left;"> LR </td> <td style="text-align:left;"> 0.8972 (0.0105) </td> <td style="text-align:left;"> 0.8241 (0.0128) </td> <td style="text-align:left;"> 0.8579 (0.0118) </td> <td style="text-align:left;"> 0.7699 (0.0142) </td> </tr> <tr> <td style="text-align:left;"> RF </td> <td style="text-align:left;"> 0.8984 (0.0110) </td> <td style="text-align:left;"> 0.8422 (0.0123) </td> <td style="text-align:left;"> 0.8985 (0.0102) </td> <td style="text-align:left;"> 0.7522 (0.0145) </td> </tr> <tr> <td style="text-align:left;"> GB </td> <td style="text-align:left;"> 0.9016 (0.0102) </td> <td style="text-align:left;"> 0.8241 (0.0128) </td> <td style="text-align:left;"> 0.8764 (0.0111) </td> <td style="text-align:left;"> 0.7404 (0.0148) </td> </tr> <tr> <td style="text-align:left;"> SVM </td> <td style="text-align:left;"> 0.8994 (0.0106) </td> <td style="text-align:left;"> 0.8331 (0.0126) </td> <td style="text-align:left;"> 0.8782 (0.0110) </td> <td style="text-align:left;"> 0.7611 (0.0144) </td> </tr> <tr> <td style="text-align:left;"> NN </td> <td style="text-align:left;"> 0.9017 (0.0103) </td> <td style="text-align:left;"> 0.8343 (0.0125) </td> <td style="text-align:left;"> 0.8542 (0.0119) </td> <td style="text-align:left;"> 0.8024 (0.0134) </td> </tr> </tbody> <tfoot><tr><td style="padding: 0; " colspan="100%"> <sup></sup> Note: Standard errors are included in parentheses.</td></tr></tfoot> </table> <br> **Final LR model includes these variables:** .small[ - State - Sex - Race - Age Group - Manner of Death ] <div style="position: absolute; bottom: 10px; right: 20px; font-size: 0.8em; color: gray;"> A - 5 </div>