Health Outcomes in the Short and Long Run
Saturday, Jan. 4, 2020 12:30 PM - 2:00 PM (PDT)
- Chair: Sabrina Wulff Pabilonia, U.S. Bureau of Labor Statistics
Comparing the PRA Program to Other Housing Options for People with Disabilities
Abstract: In 2010, Congress introduced the Project Rental Assistance (PRA) program as a reform to the existing Section 811 Supportive Housing for Persons with Disabilities Program. The goal of PRA is to provide cost-effective, affordable housing for non-elderly persons with disabilities that is integrated in multifamily developments serving people with and without disabilities, while linking residents to supportive services through partnerships between state housing agencies and state health and human services agencies. We compare outcomes (using a double-robust evaluation design) for PRA residents to those in the legacy Project Rental Assistance Contract (PRAC) program, residents in other HUD programs not linked to supportive services, and individuals unassisted by HUD subsidies. PRA units tended to be in larger buildings, two thirds of which had for-profit owners (fewer PRA buildings were owner-occupied than any comparison unit type, and more had 50 or more units per building). PRA buildings had smaller fractions of people with disabilities, on average, and were in neighborhoods with higher poverty and racial concentration, but better transit and walkability scores. PRA residents were in substantially worse health prior to move-in, judging by prevalence of diagnoses and health care utilization, but had comparable utilization after move-in. The only significant differences were higher use of personal care assistants and case management, but lower use of long-term care. Accounting for costs across multiple programs is complicated by many sources of capital funding, limited data around healthcare and services costs, and distinguishing the start-up elements of administrative costs. Costs to the government of housing and healthcare and services appear to be similar for PRA and PRAC residents. On net, PRA seems to have achieved its goals of more integrated housing options with greater case management and lower acute or long-term care, at similar costs to legacy programs.
Health Outcomes in Mid-Ages: Multistate Time to Event Statistical Models Versus Long Short Term Memory (LSTM) Recurrent Neural Network (RNN) Models
AbstractHealth outcomes such as serious diseases and death develop sequentially over one’s life span. Genetic and epigenetic factors, and health related behaviors throughout one’s life condition those health outcomes. Some individuals develop one or more diseases and their health deteriorate faster, worsening their quality of lives and survival probabilities. The models in health economics literature formulate progression of health outcomes over the life span using the multi-state time to event framework, and estimate the effects of the above mitigating factors on probabilities of transitions from one health state to another. The sequential nature of health progression is captured in the Markovian structure. Markov chain models have short memory, i.e., these models assume that given the current health outcome, the past does not influence the probability of transition to another health outcome. Many chronic diseases such as cancers and heart diseases manifest as a result of long lagged past health behaviors. Markov chain models are limited in capturing these effects. More recently, long short term memory (LSTM) recurrent neural network (RNN) models are developed in the machine (deep) learning literature that keep track of important features from the past inputs in its memory cells, which the model determine important in the ‘context’ of the future outcomes. These models are generally applied to natural language processing, creating audio streams, and video subtitling. In this paper, I adapt the existing LSTM-RNN models for the prediction of sequential health outcomes in mid-ages. I first compare the merits and shortcomings of these two approaches and then use the Health and Retirement Study (HRS) data to compare their performances in predicting sequential health outcomes.
- I1 - Health
- I3 - Welfare, Well-Being, and Poverty