USING DATA SCIENCE TO GENERATE PSYCHOSOCIAL PROFILES OF FINANCIAL EXPLOITATION IN SENIORS

Abstract Financial exploitation (FE) in older adults is a significant public health problem linked to outcomes including depression, financial ruin and early mortality. This study applied exploratory data science techniques to a multi-year statewide protective services dataset of over 8,000 elder abuse cases. The goal was to derive data-driven psychosocial profiles of abuse with an emphasis on determining which factors, commonly shared across abuse cases, were most important for determining when elder FE was occurring and whether it was occurring alone or in conjunction with other types of abuse. We found that pronounced psychological distress (i.e. verbalizing suicide, homicide, self-harm) was most important for indicating when abuse had occurred and predicted non-FE related abuse. Drug paraphernalia in the home and perpetrator drug/alcohol use were important predictors of FE-related abuse. When differentiating pure FE from hybrid FE, factors indicative of long-term FE occurrence and substantial financial loss were most important (i.e. facing foreclosure, lack of food, medications, and utilities). The findings parallel some existing work characterizing pure and hybrid FE, but also highlight new profile factors that may help determine when FE is occurring and when it is less likely. Applying data science approaches to other large protective service datasets and national datasets such as the National Adult Maltreatment Registry could help improve characterization of abuse types such as pure and hybrid FE resulting in better detection, response and prevention.


TECHNOLOGICAL AND FINANCIAL PREDICTORS OF FEAR OF FINANCIAL ABUSE AMONG OLDER ADULTS
Shaina Alves, 1 and Erin Grinshteyn 1 , 1. University of San Francisco, San Francisco, California, United States Much research has focused on elder abuse. Less research focuses on fear of abuse. This analysis examines the associations between feelings of technological competence and variables assessing financial confidence with fear of financial victimization. Data were collected among community dwelling older adults in Nevada (n=467). Questions were asked regarding technological competence, confidence in navigating the financial system, asset protection, trust in financial institutions, and previous financial abuse victimization. The outcome was assessed by asking how afraid the respondent was of becoming a victim of financial abuse. Multivariate logistic regression models were run controlling for confounding. Controlling for all covariates, those who reported feeling unconfident in their technological competence had 2.5 times the odds of being afraid of financial abuse compared with those who felt confident (p<0.02). Those who reported feeling like their assets were at risk had 4.12 times the odds of being afraid of financial abuse (p<0.0001). Older adults who reported feeling vulnerable to financial victimization had 9.4 times the odds of being afraid of financial abuse compared with those who felt invulnerable (p<0.0001). Those who were previously victims of financial abuse had 4.33 times odds of being afraid of financial abuse compared with those who had no history of financial abuse (p<0.0001). Feeling confident in the financial system, asset protection, fear of credit card use, and trust in financial institutions were not associated with fear of financial abuse. These data provide a better understanding of fear of financial abuse, which will allow for better prevention of this issue. Financial exploitation (FE) in older adults is a significant public health problem linked to outcomes including depression, financial ruin and early mortality. This study applied exploratory data science techniques to a multi-year statewide protective services dataset of over 8,000 elder abuse cases. The goal was to derive data-driven psychosocial profiles of abuse with an emphasis on determining which factors, commonly shared across abuse cases, were most important for determining when elder FE was occurring and whether it was occurring alone or in conjunction with other types of abuse.

USING DATA SCIENCE TO GENERATE PSYCHOSOCIAL PROFILES OF FINANCIAL EXPLOITATION IN SENIORS
We found that pronounced psychological distress (i.e. verbalizing suicide, homicide, self-harm) was most important for indicating when abuse had occurred and predicted non-FE related abuse. Drug paraphernalia in the home and perpetrator drug/alcohol use were important predictors of FE-related abuse. When differentiating pure FE from hybrid FE, factors indicative of long-term FE occurrence and substantial financial loss were most important (i.e. facing foreclosure, lack of food, medications, and utilities). The findings parallel some existing work characterizing pure and hybrid FE, but also highlight new profile factors that may help determine when FE is occurring and when it is less likely. Applying data science approaches to other large protective service datasets and national datasets such as the National Adult Maltreatment Registry could help improve characterization of abuse types such as pure and hybrid FE resulting in better detection, response and prevention.

PERPETRATOR DEMOGRAPHICS ASSOCIATED WITH ADULT PROTECTIVE SERVICES-CONFIRMED FINANCIAL EXPLOITATION CASES
Cassandra J. Enzler, 1 Carlos Reyes-Ortiz, 1 and Jason Burnett 1 ,

University of Texas Health McGovern Medical School, Houston, Texas, United States
Financial exploitation (FE) in older adults is a significant public health problem linked to outcomes including depression, financial ruin and early mortality. Studies have demonstrated risk factors associated with FE, but less is known about perpetrator characteristics. We performed a secondary data analysis of over 16,000 reported cases of FE utilizing a cross sectional design. Using multivariate logistic regression, confirmed and unconfirmed cases of FE were predicted from the following perpetrator demographics: age, gender, marital status, ethnicity, relationship to the victim, living status, and histories of drug abuse, alcohol abuse, and mental illness. Significant perpetrator demographics predicting confirmed FE were separation/ divorce (OR=1.48), identifying as White (OR=1.33) or Black (OR=1.44), being a daughter (OR=1.61), son (OR=1.75), grandchild (OR=2.72), or other family member (OR=1.41), not residing with the victim (OR=2.32), and having a history of drug abuse (OR=2.56), alcohol abuse (OR=1.80), or mental illness (OR=1.91). These findings are based on a large statewide dataset and describe important perpetrator characteristics that could potentially be targeted for both intervention and prevention programs. This is especially important as many victims are reluctant to seek criminal action against a family member or trusted individual. This information is valuable as it may help APS, who has limited funding and staff, investigate and intervene in more difficult elder abuse FE cases.

POLICIES AND PREVENTION OF U.S. WOMEN'S VIOLENT DEATH ACROSS AGES
Sonia L. Salari, 1 and Carrie Sillito 1 , 1. University of Utah, Salt Lake City, Utah, United States U.S. violent death rates (homicide and suicide) are the highest in the developed world. Of all female murders (femicide), the majority are male perpetrated, intimate partner violence (IPV-55-63%). Men are more often killed and by other male acquaintances, with only 2.8% IPV. Proportionally, older women (50+) have the top homicide victim rate (26%) among women. The baby boom cohort has 478 Innovation in Aging, 2019, Vol. 3, No. S1