Acoustic syndromic surveillance may be the best way to stop the spread of infectious illnesses.
In this article:
- What Is Syndromic Surveillance?
- How is Syndromic Surveillance Data Collected?
- Using Sound for Syndromic Surveillance
- Acoustic Surveillance | Walk-Through
- Acoustic Surveillance | More Than Just Coughs Per Day
Sound As Syndromic Surveillance
Syndromic surveillance collects and analyzes health indicators in real-time. It may help identify anomalies faster than traditional disease reporting allows. Identifying a health threat sooner can lead to more effective public health actions.
From an epidemiological perspective, a single cough, from one person, means very little. After all, it’s not uncommon to cough as many as dozens of times per day. Because cough is so common, even among the healthy, it doesn’t raise eyebrows.
To clarify, a cough in the cinema or classroom doesn’t worry you. But, hearing 15 or 20 coughs, from multiple people, during the opening credits of a film is concerning.
Infectious respiratory disease symptoms nearly always appear before diagnosis. And, depending on factors such as access to health care, symptoms may occur without a diagnosis.
Therefore, monitoring symptoms instead of waiting for a diagnosis is crucial. Health tracking can help indicate where and when an outbreak is occurring before medical professionals report diseases.
As we’ve learned with the COVID-19 crisis, identifying clusters of illness, is essential to mobilizing a rapid response. Which, in turn, prevents further infection.
Data for syndromic surveillance is often administrative, such as:
- School absenteeism logs.
- The register of symptoms from an emergency room’s triage nurse.
- Frequency of specific laboratory tests.
Additionally, syndromic surveillance sources are often biased. For instance, it relies on access to healthcare and geographic reach, which doesn’t fully represent the population.
And, while it’s faster than traditional disease reporting, it’s not quite as fast as disease contagion.
Someone might be sick or contagious for a while before they miss work or school. The same goes for emergency room visits.
What’s more, it only covers areas where data is prevalent and timely. Areas without emergency rooms, digitized school logs, or centralized laboratory tests registers can’t effectively carry out syndromic surveillance.
An effective syndromic surveillance system needs to be fast and cover a wide area. It should also monitor symptoms as they emerge. And, because geographic clustering is crucial to disease outbreaks, it should consider symptoms trends over time, as well as space.
Finally, an effective system needs consistent and interoperable data sources to compare areas.
We think we can use sound. More specifically, monitoring cough frequency over time and space. Tracking cough frequency may produce meaningful, real-time, actionable data on the health of a population.
We call this acoustic surveillance, which entails the following:
- Listening to sounds.
- Separating coughs from non-coughs.
- Combining and analyzing the data to detect abnormal trends.
- Taking public health action if cough frequency significantly increases in an area.
More importantly, acoustic surveillance tools already exist and are widely deployed. Most smartphones can register sound, and machine learning algorithms can distinguish between coughs and non-coughs.
We only have to turn on the system, get permission to monitor someone’s coughs, and analyze that data.
Consider Mary. She uses a cough-tracking app on her phone, and she’s familiar with her cough baseline. On average, she coughs about ten times a day.
For several days in early March, she was coughing at more than double her typical rate. Because she knows her baseline, it’s easy to recognize a cold or flu.
While we might find it interesting, public health authorities don’t. They’ve got thousands of Mary’s to monitor, and they’re relatively uninterested in one person’s case of the sniffles.
So what if instead of just looking at Mary, we looked at, say, a dozen people.
More people means a great deal of variance. There are some chronic coughers, some infrequent coughers, and also a few cases of spikes in cough, like Mary’s. But what’s important to notice is that Mary’s early-March cough-spike wasn’t part of a trend. On the whole, the disease trend for this small population remained constant.
A visualization like above tends to be too noisy. In a real syndromic surveillance system, there’s no reason to monitor every individual, every day. We compile data to get a better understanding of overall trends.
Below we see the average daily coughs in the population (black), a smoothed trend (orange), and a range of normalcy (blue).
Of course, trends over time only cover one dimension. There’s also space. And no syndromic surveillance system would be complete without a geographic component.
Let’s take a hypothetical territory, with eight districts, as an example. Each point is a household (1,000 in total).
For the sake of this example, we’ll assume it’s a rural, developing area with relatively even distribution of population and low access to healthcare services. Hence, it’s hard to carry out syndromic surveillance via practitioners.
Let’s also assume that 50% of households have a smartphone, and 10% of those have the cough-tracking app installed. So, we can track 100 households’ coughs (red).
With acoustic surveillance, we can monitor coughs over both time and space. The points below reflect the number of daily coughs for each household.
Again, this level of detail is too much for an effective acoustic syndromic surveillance system. From a public health perspective, we’re unconcerned about minor variations over time, or differences in cough frequency between specific households.
After all, a smoker’s household might have a significantly different cough pattern than that of a non-smoker. Similarly, age, phone use, device quirks, and household size might explain most of the variance in cough frequency.
Here are household-specific cough frequencies, grouped by districts. Each black line is a household, and the blue line is a smoothed trend.
Rather than household-specific trends, we’re more interested in the big picture: detecting anomalies. We can establish area-specific baseline ranges based on the previously observed data. When a cough frequency for an area exceeds the expected range, our acoustic surveillance picks up on the change and triggers an alert.
Below is the average number of household coughs (black dots) and a smoothed (blue) trend line for each district. The orange indicates the normal range, based on the previous month’s cough.
Health practitioners monitoring the data would take note of the abnormal trends in district one. Next, they might dispatch a team to investigate.
As you can see from the example above, tracking the average number of daily coughs may be useful for syndromic surveillance at a population level. However, it’s just the beginning.
A comprehensive acoustic surveillance system would both monitor coughs and learn from them. The same data that triggers the anomaly alert would also form part of a feedback loop. With time, the simple alert system could take into account factors such as the distribution of coughs throughout the day.
For instance, it might consider night-time coughing vs. day-time coughing. Or, coughing fits (dispersion) vs. uniform coughs distribution.
Lastly, with more people using the app, the surveillance system would improve. Not only the accuracy and timeliness but also in terms of more defined smaller catchment areas.
Have you tried health tracking apps? What was your experience? Share your thoughts in the comment section below!