An app that saves children's lives

In remote villages in Peru, pneumonia is one of the most common causes of death in children. Now researchers from ETH Zurich are in the process of developing an app to help identify the illness early on.

Enlarged view: Children are examined with the help of the app
A research assistant from the Environment-Health-Development-Platform in San Marcos examines a little patient with a pulse oximeter connected to a tablet. (Image: Walter Karlen / ETH Zurich)

Each morning, at 6 a. m., the same scene plays out in the small town of San Marcos in northern Peru as seven research assistants climb onto their motorcycles and set off over steep, unsurfaced roads. Their daily task: to survey the tiny villages that lie scattered around San Marco at altitudes of between 2,300 and 3,900 metres, visiting families with children who are currently taking part in a study.

The study is being conducted by researchers from ETH along with the Swiss Tropical and Public Health Institute (Swiss TPH) and Universidad Peruana Cayetano Heredia. “Our goal is to improve the living conditions and healthcare for children in this area,” says Walter Karlen, Assistant Professor and Head of ETH’s Mobile Health Systems Lab. To achieve this, Karlen is turning to mobile health technology and developing an app designed to provide a faster and more reliable way of diagnosing pneumonia in children.

Lack of early identification

Alongside diarrhoea, pneumonia is the most common illness in small children in the Cajamarca region. Many people here live in extreme poverty. The floors of the houses are made of compacted clay, kitchens have no running water and most people still cook over an open fire.

“The smoke and poor hygienic conditions help respiratory illnesses take hold,” says Daniel Mäusezahl, epidemiologist at the Swiss TPH. For years now, he has led multiple studies focusing on environmental health in the region. These aim to improve children’s health through the installation of enclosed stoves and instruction in hand washing and kitchen hygiene – and thereby to avoid illnesses, and pneumonia in particular, in the first place. According to the World Health Organisation (WHO), almost two million children worldwide die of pneumonia every year – more than malaria, AIDS and measles combined. “Often the illness is not detected early enough,” says Karlen, who is heading up the study along with Mäusezahl. The reason for this is the lack of early warning systems as well as lack of medical care. The only hospital in the region is in San Marcos – which is more than 50 kilometres away from some of the villages. There are around 25 small health centres in the catchment area, but these have no doctors and are staffed exclusively by nurses. Because it is difficult for the nurses to differentiate pneumonia from ordinary bronchitis, children are often taken to hospital only when their condition is critical. “By then, sometimes it is already too late,” says Karlen.

A new app aims to prevent cases like this by assisting medical staff as they diagnose pneumonia. An initial version is already being used as part of the study by the local research assistants, whom he trained himself. Each week, the research assistants visit each of the 300 or so families involved in the study, asking parents about the health of their children, who are all two years old or younger. They also measure the children’s breathing rate using the app. “This is an important diagnostic marker for identifying pneumonia,” says Karlen. In small children, a normal breathing rate is between 20 and 40 breaths a minute. This is usually measured by observing how many times the child’s chest rises and falls in the space of a minute. “However, people often lose count,” says Karlen, potentially leading to an inaccurate value.

More diagnostic certainty

The app eliminates this problem by taking care of the counting. All the user needs to do is tap the screen each time they see the child breathe. If there is any break in the pattern, the app counts it as an accidentally missed measurement point and has the user keep tapping until the app can determine a reliable average value. It is also possible to store the results on various days and compare them at a later date. “This provides the person taking the measurement with added certainty and supports them in making a diagnosis,” says Karlen. Previous studies have already indicated that the app is faster and just as reliable as the conventional counting technique.

Enlarged view: People siting at a table with research equipement.
Local research assistants discuss using the app during a training session. (Image: Daniel Mäusezahl / ETH Zurich)

However, the use of the app as a diagnostic tool is just one part of the study, and the priority is now on expanding the app functions. As Karlen points out: “The more indicators we can collect in addition to the breathing rate, the more reliable the diagnosis.” One of these indicators is the oxygen saturation in the blood. If this value is extremely low, it is a good indicator of severe pneumonia. For this reason, each time the research assistants also take a measurement with an instrument known as a pulse oximeter, a standard piece of medical kit in these parts. A sensor attached to the finger uses infrared light to measure the levels of oxygen in the blood.

The study is intended to demonstrate whether or not blood oxygenation is also suitable for use as a diagnostic criterion within the app. To do this, the researchers have already developed a mathematical model that factors in the oxygen levels of the air at a given altitude. It then correlates this with blood oxygen levels, which allows it to determine the probability of severe pneumonia in the patient. However, there are still a few hurdles to overcome before this becomes a viable method. One is that blood oxygen levels vary widely from individual to individual, even in healthy people. There is also the fact that oxygen levels vary with altitude: as the air becomes thinner, blood oxygen levels decline. This makes it difficult to determine standard values, particularly in the case of children living high in the Andes.

For the moment, the researchers want to collect as much data as possible from a range of altitudes so that they can validate their model. Since the research assistants also note each child’s state of health on each visit, it is also possible to compare values from healthy and sick children.

If the results are promising, the researchers intend to create a new, expanded app. This could then be rolled out to the local health centres in collaboration with local health staff, and tested for everyday practicability. “Apps have enormous potential to improve healthcare in places lacking medical equipment and expertise,” says Karlen. After all, these days smartphones and tablets are cheap and commonly found even in low and middle income countries. Karlen hopes that his app will be one of those helping to save children’s lives in the future.

This article appears in the current issue of Globe.

JavaScript has been disabled in your browser