This is the first study evaluating the use of wearable biosensor devices for the continuous physiological monitoring of individuals with acute Lassa fever. Although this study was not powered to make conclusions about physiological parameters and their associations with clinical outcomes, we did observe that all but one patient who did not survive their illness had a mean HR in the age-specific tachycardic range, compared to none of the survivors. These findings are similar to what has been observed in similar settings for patients with bacterial sepsis²¹. Conversely, we found a less pronounced difference in mean RR between groups. It is worth noting that four of the five non-survivors were pediatric patients, while only one of the three survivors was a child, who are known to have higher resting HR and RR than adults¹⁹. We attempted to account for this by using established age-specific thresholds to define tachycardia and tachypnea. Elevations in respiratory rate during Lassa fever could be explained by several potential etiologies including encephalopathy, anxiety, hypoxemia, pleural effusion, acute anemia, lactic acidosis from impaired tissue perfusion, or acidosis from renal failure^22,23.

There is robust clinical evidence that alterations in HRV correlate with severity of systemic infection and have prognostic value as a predictor of multiorgan dysfunction and death^11,24,25,26. A normal HRV is between 19 and 75 ms. Though normal ranges for children have not been firmly established, it is believed that there is a linear relationship between HRV and age²⁷. On average, individuals who survived acute Lassa fever demonstrated HRV within the normal range (mean HRV 59 ms), while those who died were characterized by HRV below the normal range (mean HRV 10 ms)²⁸. Increases in HRV over the course of infection have also been shown to correlate with survival in ICU patients with septic shock²⁹. In our study, non-survivors failed to mount any sustained increases in HRV. Pre-admission measurements for all patients were unavailable, which precluded the opportunity for comparison of HRV during acute illness with individuals’ healthy baselines. Still, results from our small sample suggest there may be a role for the assessment of HRV in Lassa fever prognostication that can be defined with further investigation.

The shaded gray areas in Fig. 4 represent periods of patient decompensation toward death that were captured by objective changes in vital signs, such as an acute increase in HR or RR. If used as real-time telemetry, the study’s monitoring system has the potential to identify these changes more rapidly than the current system of collecting patient vital signs every 30 min at the KGH Lassa ward, creating opportunities for earlier intervention and better outcomes⁷. This benefit may be even more pronounced in other regions of Lassa-endemic West Africa, where in-person patient monitoring by nursing is often more limited.

Further adoption of wearable biosensor devices to remotely monitor individuals with severe acute Lassa fever poses several challenges. First, we encountered recurrent lapses in patient data collection due to failure of the devices’ adhesive backing, which was indicated by elevations in impedance detected by the sensor. Temperatures during the dry season in Sierra Leone frequently surpass 38 °C with a high relative humidity³⁰. These climatic conditions, along with poor ventilation in patient rooms in the Lassa ward, likely led to patient perspiration causing occasional adhesive failure on the devices.

Second, due to the floor plan of the Lassa ward at KGH, Android devices were occasionally out of range for the monitoring devices’ Bluetooth connectivity, leading to further lapses in patient data collection. This was addressed by situating the Android devices as close as possible to the patient area, though they could ultimately not be brought into the patient area due to the location of power outlets and concerns about infection control. However, even when both devices were within the Bluetooth radius of 10 meters, there were still occasional connectivity problems, an issue that was present in a previous study conducted in a Rwandan emergency department⁷.

Third, we observed several spurious vital sign readings that would be physiologically impossible, such as very extreme skin temperatures or respiratory rates of zero. These readings prompted us to discard patient data that was generated during the same 1-min window and raised concerns about rare but easily identifiable issues with device calibration. To explore whether profound hemodynamic shock may have affected skin temperature readings in our three adult participants, we reviewed manual vital signs at intervals when biosensor-reported skin temperatures were non-physiological. We found no temporal correlation between abnormal skin temperature reported by the biosensor and blood pressure as measured manually. We were unable to perform this same analysis for our pediatric participants, as there was no access to a pediatric sphygmomanometer at the KGH Lassa ward during the study period, and manual blood pressures were not collected for pediatric patients.

Fourth, at nearly US$170.00 each in 2024, the cost of wearable biosensor devices–not including the necessary Android devices to accompany them–may be prohibitive for many settings in which Lassa fever is endemic. Fifth, the VitalPatch device is intended for use on general care patients who are 18 years of age or older³¹. Although the device appears capable of collecting high-quality data from young children–as evidenced by the presence of 100% EKG signal quality from participants in our study as young as five months–we encountered issues with pediatric participants removing the device and disrupting the collection of data.

Lastly, Sierra Leone suffers from a profound shortage of clinicians, with only one physician per ten thousand population³². The country’s official Lassa ward does not have access to organ support equipment, such as mechanical ventilators or hemodialysis machines, and routine laboratory studies are performed infrequently. Even with the ability to identify patients at risk for death based on vital sign trends or acute decompensation, the burden on local physicians and the extreme scarcity of sufficient resources for meaningful intervention limits this technology from being implemented to its full potential.

Our study has several limitations. A more robust statistical analysis of our patients was limited by a small sample size, owing to the relative scarcity of diagnosed Lassa fever in Sierra Leone, which averaged just 8.4 cases annually from 2019 to 2023 (unpublished; from KGH records). Additionally, the majority (81.3%) of continuous physiological data we collected was ultimately discarded due to poor quality. There was also a bias toward younger individuals in our study, with the median age falling in the pediatric range. This is in accordance with existing epidemiologic data from Sierra Leone that suggest the highest incidence of antigenemic Lassa fever is observed in children and young adults¹. The distribution of ages between survivors and non-survivors in our cohort limits the conclusions we can draw from our vital sign data, as younger patients on average have increased HR, increased RR, and decreased HRV compared to adults. Given our standard of 12 h of continuous high-quality data, participants who died in less than 12 h were not included in the analysis, which biased our results away from more imminently ill patients.