In almost all communities in the United States, the number of youth experiencing homelessness exceeds the capacity of housing resources available to youth. This situation leaves communities with the terrible predicament of trying to decide whom to prioritize for the precious few housing spots available at any given time. For adults, this same dynamic exists and many communities have turned to vulnerability assessment tools to help them make these difficult decisions. Most of these tools have focused on assessing factors that are associated with either premature mortality or greatest system costs. Youth aged 24 or younger are unlikely to experience health-related premature mortality, nor are they likely to have incurred enormous system costs. Thus in recent years a new set of tools was developed that specifically targets the needs and realities of homeless youth. The two most prominent of those tools are the TAY Triage Tool, developed by the Corporation for Supportive Housing (CSH) and Dr. Eric Rice, and the Next Step Tool for Homeless Youth, which was developed by Orgcode in consultation with CSH and Dr. Rice.
The TAY Triage Tool is a short 7-item (6-point) index based on extensive research Dr. Rice conducted in conjunction with CSH. Unlike the adult tools, which are based on developing predictors of system cost or premature mortality, the TAY Triage Tool is anchored in assessing which youth are most likely to experience long-term homelessness. This tool was created using standard statistical approaches to “predictive” modeling, examining associations with long term homelessness and systematically reducing the number of variables used in logistic regression models to come to a final model with 6 independent “predictors” of long term homelessness.
Iain De Jonge and Orgcode were responsible for the creation of the Next Step Tool for Homeless Youth (Orgcode, 2015). They have described it as an evidence-informed tool, because its creation is based primarily on an extensive review of the scientific literature on vulnerability factors for homeless youth. In addition, in consultation with CSH and Dr. Rice, they incorporated items from the TAY Triage Tool into their larger assessment tool. It is worth noting that we eliminated some items they used from the TAY Triage Tool because they did not differentiate between individuals who had experienced longer-term homelessness and those who had not. For example, one excluded item was: “Have you been attacked or beaten up since you’ve become homeless?” Conversely, the Next Step Tool incorporated several items we did not consider, such as “Does anybody force or trick you to do things that you do not want to do?” Neither tool is perfect. Both, however, strive to identify vulnerable youth and help communities prioritize housing for youth based on objective criteria known to assess vulnerability.
The challenge we face now is can we use techniques from artificial intelligence to generate superior predictive models that could help communities more effectively identify and prioritize youth for intensive housing interventions.