Browsing by Subject "risk assessment"

Now showing 1 - 4 of 4

Algorithmic Risk Assessments in the Hands of Humans
(Salem Center, 2020-10-19) Doleac, Jennifer L.; Stevenson, Megan T.
COVID-19 risk assessment for public events - May 2022
(2022-05) Fox, Spencer J.; Johnson, Kaitlyn; Owirodu, Briana; Johnson-Leung, Jennifer; Elizondo, Marcel; Walkes, Desmar; Meyers, Lauren Ancel
We describe a risk assessment framework to support event planning during COVID-19 waves. The method was developed in partnership with public health officials in Austin, Texas. The framework is based on a previously published model [1]. The inputs to our calculations include the following: the local prevalence of COVID-19 [2], epidemiological properties of current variants, the structure of the event, including the number of attendees, types and duration of activities, density of interactions, and ventilation, COVID-related precautions for the event, including vaccine, testing, and face mask requirements, and local demographic information. The risk assessment framework uses the above inputs to estimate the following quantities: the number of attendees likely to arrive infected, the reproduction number of COVID-19 at the event, the number of attendees likely to become infected at the event, and the number of additional infections that will occur in Austin in the subsequent four weeks, stemming from infections occurring at the event. This report considers two case studies in Travis county (Austin, TX): (1) a business conference with 3,000 attendees and (2) an outdoor festival with 50,000 attendees.
Hazards Analysis, Building 950, Brooks Air Force Base, Bexar County, Texas
(1975) Chin, John L.; Jones, Leslie
Based on available information, building 950 is a suitable site for storage of lunar samples. The only apparent hazard is the high shrink-swell capacity of the soil. This soil property should be a primary consideration in any modifications made to the building.
Regulating Algorithmic Pretrial Risk Assessment: Lessons from Texas Bail Reform Efforts
(2020-05) Ashok, Arvind
Algorithmic risk assessment is increasingly used to gauge the “risk” of defendants in a pretrial context. Given the noted socioeconomic disparities and due process issues with the current pretrial system, which frequently relies on monetary bail, algorithmic risk assessments may help judges come to more rational decisions on whether to detain defendants before trial. However, these assessments have also been critiqued for alleged bias, lack of transparency, and a false sense of objectivity. If algorithmic risk assessments continue, careful regulation of their use is likely necessary, but currently lacking. This thesis synthesizes existing literature on pretrial detention, risk assessment, and algorithmic accountability to identify the potential and flaws of algorithmic risk assessment within all the relevant contexts. Then, it provides a case study of the adoption of an algorithmic pretrial risk assessment tool in Texas in order to elucidate how real-world administrators are managing risk assessments and responding to concerns raised in the literature. The finding from both the synthesis and the case study is that better institutional design and active regulation are needed to keep risk assessments democratically accountable, useful to the broader goals of bail reform, and free from damaging politicization. To this end, the final section of this thesis proposes best practices for future administrators of algorithmic risk assessment, taking into account lessons from both Texas and the broader academic literature.