Results of a scientific study by Bambin Gesù Hospital in Rome and telehealth startup Paginemediche have been published in the Journal of Medical Internet Research
The use of digital medicine platforms and anonymous user data collected online can help predict the spread of an infectious disease, identify geographical locations of higher or lower prevalence, and support pandemic responders.
This insight emerged from a study recently published in the Journal of Medical Internet Research (JMIR) led by digital experts in collaboration with the Predictive and Preventive Medicine Research Unit of the Bambin Gesù Hospital in Rome, the largest pediatric hospital and research center in Europe, and Paginemediche, an innovative digital medicine startup.
During a pandemic, users accessing a digital system are a valuable source of information that can facilitate traditional surveillance activities, allow for earlier prediction of a spike in the incidence of disease, and support preventive strategy decisions. For example, the Ministry of Health used the data from positive COVID-19 swabs along with other indicators to plan preventive strategies, including restrictions on travel and social activities, and to predict possible outcome scenarios.
As disease surveillance is affected by the time needed to make a diagnosis and communicate the associated data, epidemiologists are interested in evaluating information complementary to traditional surveillance that can provide early predictions. However, the availability of rapid information has to be weighed against the reliability of that information. As a result, internet user data has often been regarded as ancillary to disease surveillance; an area giants such as Google have explored with varying success.
The team of researchers examined a total of 75,557 sessions in the online decision support system (chatbot) developed by Paginemediche, a simple tool available that aims to answer user questions about COVID-19 and recommend the most appropriate behavior in accordance with the Ministry of Health.
The recommendations particularly concerned users with symptoms or those in close contact with a COVID-19 positive person. This user decision-support system, freely accessible in an algorithm-driven chat room, has been in place since the start of the pandemic in March 2020 across the country and has now been extended to assess other conditions and support early identification of additional diseases beyond COVID-19.
The study was conducted by accessing data from Paginemediche’s assisted decision-making system and comparing it with surveillance data distributed by the Ministry of Health in order to assess the degree of concordance over time. Although the assisted decision-making system could not accurately predict the number of cases that were notified to the Ministry of Health, it was able to predict the upward or downward trend of cases across the country one week in advance of the Ministry of Health’s surveillance data. The accuracy in anticipating pandemic trends was better when it considered users who had been in contact with a patient positive for COVID-19.
Specifically, 65,207 sessions were recorded from users with symptoms, 19,062 from contacts with individuals with COVID-19. The highest number of sessions in the online decision support system was recorded in the early stages of the pandemic. A second peak was observed in October 2020 and a third peak was observed in March 2021, in parallel with the wave of reported cases. The online decision support system session peaks preceded the wave of reported COVID-19 cases by approximately one week.
The results of the study are consistent with the consideration that awareness of contact with a positive individual or having respiratory symptoms anticipates the possible diagnosis by nasal swab and subsequent notification to the Ministry of Health by a few days. Although data from an open, uncontrolled system may fluctuate for different, unpredictable reasons and are not as robust as those based on laboratory tests, these systems represent a source of information that can complement traditional surveillance activities, allow for earlier prediction of possible increases in disease cases, and support decisions for preventive strategies by public health institutions.
