Harvard OpenMetabolics Turns Smartphones Into Powerful Health Tools

A basic smartphone carried in a pocket may soon serve as a highly capable health device. Researchers at Harvard University have created a system that calculates the energy a person expends in everyday activities, and it could be considerably more precise than many widely used fitness watches.

Scientists from the John A. Paulson School of Engineering and Applied Sciences (SEAS) developed the smartphone-based system, called OpenMetabolics. It monitors leg movement to calculate the number of calories an individual burns.

The complete study sets out how the system operates and why it could address an important health challenge.

Measuring physical activity

Physical inactivity is the fourth leading cause of death globally. Moving regularly supports muscle strength, cardiovascular health, mental wellbeing, sleep and even brain function.

Yet researchers continue to find it difficult to establish clear links between physical activity and numerous health results, including weight loss and quality of life for people with certain illnesses.

Health bodies such as the World Health Organization have called for improved methods of measuring physical activity.

To gain a genuine understanding of health, scientists must establish how frequently someone moves, the duration of each activity and its intensity. Brief periods of walking throughout the day are important too, rather than only lengthy exercise sessions.

The problem with fitness trackers

Many smartwatches and fitness trackers calculate calorie expenditure from heart rate and wrist movements. Their estimates can contain substantial inaccuracies, with some research reporting errors of between 30 and 80 percent.

Laboratory methods, including direct calorimetry and respirometry, can assess energy expenditure with great accuracy. However, they depend on specialised equipment and are not practical for routine use in everyday life.

Questionnaires also ask people to describe their own activity levels, although faulty recollection and personal bias may produce inaccurate responses.

Smartphones could provide a more effective alternative. About 70 percent of the world's population uses a smartphone, making them more available than smartwatches across many regions.

How OpenMetabolics works

OpenMetabolics tracks leg movement rather than wrist movement. When people walk, run, climb stairs and cycle, the leg muscles consume most of the body's energy.

Monitoring leg motion therefore allows the system to estimate energy expenditure more directly. The smartphone relies on its integrated sensors, including a gyroscope and accelerometer.

The system separates movement into gait cycles, meaning a complete stepping pattern. It then applies a machine-learning model known as gradient-boosted trees to calculate the energy used in every step.

While being trained, the model received data from 36 participants undertaking activities such as walking, running, stair climbing and cycling at varying intensities. It learned the relationship between leg movement and actual energy expenditure recorded using laboratory equipment.

Forward and backward leg movement supplied the most valuable data. Height and weight had a far more limited influence on the estimates. This indicates that leg motion provides a strong indication of the energy the body is using.

Testing OpenMetabolics for accuracy

The researchers evaluated OpenMetabolics with additional participants who had not contributed to the training data. In real-world walking, it produced a cumulative error of about 13 percent.

When all real-world activities were considered together, the error was about 18 percent. This made it about twice as accurate as many commercial devices.

Participants walked outdoors on pavements, climbed stairs, ran and cycled. The study compared OpenMetabolics with a Fitbit smartwatch, a heart-rate model, a pedometer and a thigh-mounted accelerometer. OpenMetabolics achieved the lowest overall error.

The findings further indicated that age, gender and body mass index did not have a significant effect on the system's accuracy. This suggests the tool performs effectively among different groups of people.

Solving the pocket problem

One significant difficulty involved the way a phone moves inside a pocket. Loose clothing may cause a handset to shake in ways that do not reflect actual leg movement. The team addressed this by developing a pocket-motion correction model.

The correction model cut movement errors by about 28 percent. Once corrected, there was no meaningful difference between a phone secured firmly to the thigh and one carried normally in a pocket.

Users therefore do not require dedicated straps or additional equipment.

Monitoring activity for a full week

The team also assessed OpenMetabolics in a seven-day study. Participants kept a smartphone in their pocket as they went about daily life. The system measured energy expenditure for each step and revealed distinct patterns from one day to another.

For instance, activity commonly rose during commuting periods. The data also found that participants were less active on Sundays than on weekdays.

This level of detailed data could enable doctors, public-health specialists, urban planners, nutritionists and researchers to develop more effective health programmes. It could also allow scientists to examine how everyday routines influence long-term health.

OpenMetabolics, a tool for global health

OpenMetabolics is open source, meaning that researchers can access both its data and code. This gives scientists worldwide a simpler way to refine the system.

As smartphones are widely used even in underserved areas, the tool could contribute to narrowing global health inequalities.

By providing more accurate, accessible physical-activity information, OpenMetabolics may help answer major health questions and support improved decisions by individuals and communities.

A straightforward phone carried in a pocket could soon become one of the most powerful health tools available.