An AI-empowered indoor digital contact tracing system for COVID-19 outbreaks in residential care homes

An AI-empowered indoor digital contact-tracing system was developed using a centralized architecture and advanced low-energy Bluetooth technologies for indoor positioning, with careful preservation of privacy and data security. We analyzed the contact pattern data from two RCHs and investigated a COVID-19 outbreak in one study site. To evaluate the effectiveness of the system in containing outbreaks with minimal contacts under quarantine, a simulation study was conducted to compare the impact of different quarantine strategies on outbreak containment within RCHs. The significant difference in contact hours between weekdays and weekends was observed for some pairs of RCH residents and staff during the two-week data collection period. No significant difference between secondary cases and uninfected contacts was observed in a COVID-19 outbreak in terms of their demographics and contact patterns. Simulation results based on the collected contact data indicated that a threshold of accumulative contact hours one or two days prior to diagnosis of the index case could dramatically increase the efficiency of outbreak containment within RCHs by targeted isolation of the close contacts. This study demonstrated the feasibility and efficiency of employing an AI-empowered system in indoor digital contact tracing of outbreaks in RCHs in the post-pandemic era.

Clinical profile and containment of the Ebola virus disease outbreak in two large West African cities, Nigeria, July-September 2014

INTRODUCTION

The Ebola virus disease (EVD) outbreak in Nigeria began when an infected diplomat from Liberia arrived in Lagos, the most populous city in Africa, with subsequent transmission to another large city.

METHODS

First-, second-, and third-generation contacts were traced, monitored, and classified. Symptomatic contacts were managed at Ebola treatment centers as suspected, probable, and confirmed EVD cases using standard operating procedures adapted from the World Health Organization EVD guidelines. Reverse transcription PCR tests confirmed EVD. Socio-demographic, clinical, hospitalization, and outcome data of the July-September 2014 Nigeria EVD cohort were analyzed.

RESULTS

The median age of the 20 EVD cases was 33 years (interquartile range 26-62 years). More females (55%), health workers (65%), and persons <40 years old (60%) were infected than males, non-health workers, and persons aged ≥40 years. No EVD case management worker contracted the disease. Presenting symptoms were fever (85%), fatigue (70%), and diarrhea (65%). Clinical syndromes were gastroenteritis (45%), hemorrhage (30%), and encephalopathy (15%). The case-fatality rate was 40% and there was one mental health complication. The average duration from symptom onset to presentation was 3±2 days among survivors and 5±2 days for non-survivors. The mean duration from symptom onset to discharge was 15±5 days for survivors and 11±2 days for non-survivors. Mortality was higher in the older age group, males, and those presenting late.

CONCLUSION

The EVD outbreak in Nigeria was characterized by the severe febrile gastroenteritis syndrome typical of the West African outbreak, better outcomes, rapid containment, and no infection among EVD care-providers. Early case detection, an effective incident management system, and prompt case management with on-site mobilization and training of local professionals were key to the outcome.

A detection dog to identify patients with Clostridium difficile infection during a hospital outbreak

OBJECTIVES

Early and rapid identification of Clostridium difficile infections (CDI) is important to prevent transmission. In this study we assessed the diagnostic accuracy of a trained detection dog for detecting CDI cases on hospital wards in an outbreak setting.

METHODS

During a CDI outbreak in a large Dutch university hospital, we screened affected hospital wards repeatedly with a trained detection dog. The dog's response was compared to the clinical diagnosis, supported by laboratory results.

RESULTS

During a total of 9 hospital visits, the dog performed 651 screenings involving 371 participants. The dog correctly identified 12 out of 14 CDI cases [sensitivity 86% (95% confidence interval (CI): 56-97%)] and 346 out of 357 CDI negative participants [specificity of 97% (95% CI: 94-98%)]. Interestingly, of the 11 CDI negative participants that were 'falsely' indicated by the dog as positive, 2 (18%) did actually developed CDI during the 3 months of follow-up after the detection period; compared to only 12 of the 346 participants (3.5%) that the dog identified as C. difficile negative (p = 0.06).

CONCLUSION

A trained detection dog can accurately detect CDI in hospitalized patients during an outbreak. A (repeated) positive dog response is a strong indication of a CDI episode coming, be it the next day or possibly up to a month.