Predicting state level suicide fatalities in the united states with realtime data and machine learning.

Digital trace data and machine learning techniques are increasingly being adopted to predict suicide-related outcomes at the individual level; however, there is also considerable public health need for timely data about suicide trends at the population level. Although significant geographic variation in suicide rates exist by state within the United States, national systems for reporting state suicide trends typically lag by one or more years. We developed and validated a deep learning based approach to utilize real-time, state-level online (Mental Health America web-based depression screenings; Google and YouTube Search Trends), social media (Twitter), and health administrative data (National Syndromic Surveillance Program emergency department visits) to estimate weekly suicide counts in four participating states. Specifically, per state, we built a long short-term memory (LSTM) neural network model to combine signals from the real-time data sources and compared predicted values of suicide deaths from our model to observed values in the same state. Our LSTM model produced accurate estimates of state-specific suicide rates in all four states (percentage error in suicide rate of -2.768% for Utah, -2.823% for Louisiana, -3.449% for New York, and -5.323% for Colorado). Furthermore, our deep learning based approach outperformed current gold-standard baseline autoregressive models that use historical death data alone. We demonstrate an approach to incorporate signals from multiple proxy real-time data sources that can potentially provide more timely estimates of suicide trends at the state level. Timely suicide data at the state level has the potential to improve suicide prevention planning and response tailored to the needs of specific geographic communities.

A Tale of Two Cities: What's Driving the Firearm Mortality Difference in Two Large Urban Centers?

INTRODUCTION: Per police data, the case fatality rate (CFR) of firearm assault in New Orleans (NO) over the last several years ranged between 27% and 35%, compared with 18%-22% in Philadelphia. The reasons for this disparity are unknown, and potentially reflect important system differences with broader implications for the reduction of firearm mortality. METHODS: A retrospective analysis of police and city-specific trauma databases between 2012 and 2017 was performed. Victims of firearm assaults within city limits were included. Univariate analysis was performed using chi-square for categorical and t-test for continuous variables. Bivariate analysis was conducted using logistic regression. RESULTS: Per police data, the CFR of firearm assault was 31% in NO and 20% in Philadelphia. However, per trauma registry data, the CFR of firearm assault was 14% in NO and 25% in Philadelphia. Patients in Philadelphia were older, had higher injury severity score, and lower blood pressure. Patients in NO had higher rates of head injury. 51% of patients in Philadelphia arrived via police compared to <1% in NO. There was no mortality difference between police and emergency medical service (EMS) transport. Longer EMS prehospital times were associated with increased mortality in NO but not Philadelphia. A much larger percentage of patients died on-scene in NO than Philadelphia. CONCLUSIONS: Our findings suggest that the major driver of increased mortality following firearm assault in NO compared with Philadelphia is death prior to the arrival of first responders. Interventions that shorten prehospital time will likely have the greatest impact on mortality in NO. This should include the consideration of police transport.

Evidence for Limited Early Spread of COVID-19 Within the United States, January-February 2020.

From January 21 through February 23, 2020, public health agencies detected 14 U.S. cases of coronavirus disease 2019 (COVID-19), all related to travel from China (1,2). The first nontravel-related U.S. case was confirmed on February 26 in a California resident who had become ill on February 13 (3). Two days later, on February 28, a second nontravel-related case was confirmed in the state of Washington (4,5). Examination of four lines of evidence provides insight into the timing of introduction and early transmission of SARS-CoV-2, the virus that causes COVID-19, into the United States before the detection of these two cases. First, syndromic surveillance based on emergency department records from counties affected early by the pandemic did not show an increase in visits for COVID-19-like illness before February 28. Second, retrospective SARS-CoV-2 testing of approximately 11,000 respiratory specimens from several U.S. locations beginning January 1 identified no positive results before February 20. Third, analysis of viral RNA sequences from early cases suggested that a single lineage of virus imported directly or indirectly from China began circulating in the United States between January 18 and February 9, followed by several SARS-CoV-2 importations from Europe. Finally, the occurrence of three cases, one in a California resident who died on February 6, a second in another resident of the same county who died February 17, and a third in an unidentified passenger or crew member aboard a Pacific cruise ship that left San Francisco on February 11, confirms cryptic circulation of the virus by early February. These data indicate that sustained, community transmission had begun before detection of the first two nontravel-related U.S. cases, likely resulting from the importation of a single lineage of virus from China in late January or early February, followed by several importations from Europe. The widespread emergence of COVID-19 throughout the United States after February highlights the importance of robust public health systems to respond rapidly to emerging infectious threats.