Text mining have gained great momentum in recent years, with user-generated content becoming widely available. One key use is comment mining, with much attention being given to sentiment analysis and opinion mining. An essential step in the process of comment mining is text pre-processing; a step in which each linguistic term is assigned with a weight that commonly increases with its appearance in the studied text, yet is offset by the frequency of the term in the domain of interest. A common practice is to use the well-known tf-idf formula to compute these weights.
2 April 2002 in Journal of Organizational Behavior
The effect of emergency department (ED) length of stay (EDLOS) on in-hospital mortality (IHM) remains unclear. The aim of this systematic review and meta-analysis was to determine the association between EDLOS and IHM. We searched the PubMed, Medline, Embase, Web of Science, Cochrane Controlled Register of Trials, CINAHL, PsycInfo, and Scopus databases from their inception until 14–15 January 2022. We included studies reporting the association between EDLOS and IHM. A total of 11,337 references were identified, and 52 studies (total of 1,718,518 ED patients) were included in the systematic review and 33 in the meta-analysis. A statistically significant association between EDLOS and IHM was observed for EDLOS over 24 h in patients admitted to an intensive care unit (ICU) (OR = 1.396, 95% confidence interval [CI]: 1.147 to 1.701; p < 0.001, I2 = 0%) and for low EDLOS in non-ICU-admitted patients (OR = 0.583, 95% CI: 0.453 to 0.745; p < 0.001, I2 = 0%). No associations were detected for the other cut-offs. Our findings suggest that there is an association between IHM low EDLOS and EDLOS exceeding 24 h and IHM. Long stays in the ED should not be allowed and special attention should be given to patients admitted after a short stay in the ED.
There is growing agreement among researchers and developers that in certain machine-learning (ML) tasks, it may be advantageous to keep a “human in the loop” rather than rely on fully autonomous systems. Continual human involvement can mitigate machine bias and performance deterioration while enabling humans to continue learning from insights derived by ML. Yet a microlevel theory that effectively facilitates joint and continual learning in both humans and machines is still lacking. To address this need, we adopt a design science approach and build on theories of human reciprocal learning to develop an abstract configuration for reciprocal human-ML (RHML) in the context of text message classification. This configuration supports learning cycles between humans and machines who repeatedly exchange feedback regarding a classification task and adjust their knowledge representations accordingly. Our configuration is instantiated in Fusion, a novel technology artifact. Fusion is developed iteratively in two case studies of cybersecurity forums (drug trafficking and hacker attacks), in which domain experts and ML models jointly learn to classify textual messages. In the final stage, we conducted two experiments of the RHML configuration to gauge both human and machine learning processes over eight learning cycles. Generalizing our insights, we provide formal design principles for the development of systems to support RHML. This paper was accepted by D. J. Wu, Special Issue on the Human-Algorithm Connection. Funding: This work was supported by the Israel’s Ministry of Defence [Grant R4441197567] and the Israel’s Ministry of Science and Technology [Grant 207076]. Supplemental Material: The data files are available at https://doi.org/10.1287/mnsc.2022.03518 .
Rapid naloxone administration is crucial in reversing an opioid overdose. We investigated whether equipping community members, including people who use opioids (PWUO), with a smartphone application enabling them to signal and respond to suspected overdose would support naloxone administration in advance of Emrgency Medical Services (EMS). This observational cohort study of opioid overdose intervention used a dedicated smartphone app, UnityPhilly, activated by volunteers witnessing an overdose to signal other nearby volunteers in Philadelphia (March 2019 - February 2020). Alerted volunteers chose to respond, or declined to respond, or ignored/missed the alert. Witnessing volunteer was connected to 9-1-1 through a semi-automated telephone call. The primary outcome was layperson-initiated overdose reversal before EMS arrival, and a secondary outcome was hospital transfer. This study is registered with ClinicalTrials.gov, NCT03305497. 112 volunteers, including 57 PWUO and 55 community members, signaled 291 suspected opioid overdose alerts. 89 (30⸱6%) were false alarms. For 202 true alerts, the rate of layperson initiated naloxone use was 36⸱6% (74/202 cases). Most naloxone-use cases occurred in the street (58⸱11% (43/74)) and some in home settings (22⸱98% (17/74)). The first naloxone dose was provided by a nearby volunteer responding to the alert in 29⸱73% (22/74) of cases and by the signaling volunteer in 70⸱27% (52/74) of cases. Successful reversal was reported in 95⸱9% (71/74) of cases. Layperson intervention preceded EMS by 5 min or more in 59⸱5% of cases. Recovery without hospital transport was reported in 52⸱7% (39/74) of cases. Our findings support the benefits of equipping community members, potentially witnessing suspected opioid overdose, with naloxone and an emergency response community smartphone app, alerting EMS and nearby laypersons to provide additional naloxone. Funding provided by NIH through NIDA, grant number: 5R34DA044758.
1 May 2000 in IEEE Intelligent Systems and their Applications