Platform uses artificial intelligence to calculate body measurements from photographs

The solution, developed by a startup supported by FAPESP, estimates eight parameters with an error of under 5% for 80% of users, according to the firm’s founder.

By Roseli Andrion  |  Agência FAPESP – Many healthcare procedures include measurement of body weight and other anthropometric parameters, often separately by a clinician and a nutritionist, as well as a physical therapist and even an aesthetician in some cases, without sharing of the information among these professionals.

What if there were a solution that let the patient and everyone else involved verify both the data and their evolution over time? This has been proposed by Inovia Pesquisa Aplicada, a startup headquartered in Campinas (São Paulo state, Brazil) whose platform B-YOU was developed with the support of FAPESP’s Innovative Research in Small Business Program (PIPE).

Fernanda Balieiro, CEO of Inovia, had the original idea based on her own experience. “While undergoing aesthetic treatment, she looked for an app that could show her progress, but found none,” Tiago Carvalho, co-founder of the startup, told Agência FAPESP.

According to Carvalho, a specialist in artificial intelligence (AI) with computer vision, records of Balieiro’s sessions were disorganized and some were even lost. “The clinic lacked proper tools to safeguard records, which weren’t shared with patients or by the professionals involved. They used paper forms, with the professional responsible for a session taking photographs on a smartphone. In the next session, if a different professional was involved, they took other photos using their own phone. To obtain the images from the start of the treatment, you had to contact all the professionals in charge of sessions. That’s why information was often lost,” he explained.

The researchers also noticed that many clinics failed to keep track of their patients’ progress. “A patient would book sessions to reduce measurements, for example, but there was no prior or posterior control of the efficacy of the treatment,” he said.

Another difficulty observed by the team was that measurements were not standardized, making for inconsistency in data collected on the same patient by different professionals. When they spoke to professionals in the industry, they heard similar stories. “So we began thinking about a way to standardize the process,” he said.

Web-based platform

The first step taken to develop B-YOU was the creation of an app for patients to take their own measurements. “Next we decided the ideal solution would be a web-based platform for professionals in the industry to use. This is an extra benefit for patients, who can then be given access to all the data,” Carvalho said. “One of the nutritionists we interviewed said he had many patients in other countries and couldn’t find partners to take measurements in all of these locations. This is a very useful tool for him.”

In its final form, the system developed by Inovia uses AI to calculate body measurements from photos. It estimates eight parameters, with an error rate of less than 5% for 80% of the users tested so far. “If manual measurements are taken by different people, there will be differences. We’re currently improving the technology to reduce this margin of error still further, and to include other functions in accordance with the needs of the market,” he said.

The data obtained is uploaded to the platform, and users can keep track of their progress. They can also show the information to the professionals who work with them. “It’s like open banking in that individuals own their data rather than a clinic or specific professional, and they can share them with whomever they wish,” he said.

The platform will also include questionnaires produced by professionals to understand the habits of their patients. “So it will all be centralized,” he said, adding that the tool also has a motivational component. “It’s a way of keeping the user engaged. Users find it highly gratifying to see their progress in this way and feel more motivated as a result.”

The system has been evaluated by 326 users in 445 situations. “We collected feedback, and we’re correcting the platform in preparation for a second round of tests. We have beta users who are aware the product is being developed. We want to have the platform stable and ready for launch in October,” he said.

Other markets

Small online apparel retailers are another market affected by variations in these measurements. If a customer receives an item of the wrong size, the sale is unlikely to be finalized. “Customers are typically interested in an item at a specific moment. If it’s not right, they’ll return it and look elsewhere. Small e-commerce enterprises stock only a few units of a given size, so they may lose sales if sizes are wrong for customers. They may miss the chance to sell to someone else for whom the size is right,” Carvalho said.

The solution is also a good fit for the occupational health and safety market. The startup already has a partnership with a service provider in this segment. “They offer workplace exercise programs with progress tracking for participants. The platform can detect whether they’re underperforming even if they do sessions correctly, and verify whether an occupational disease injury may be involved,” he said.

Inovia plans to develop a minimum viable product (MVP) with added functionality for this service provider. “For example, it will be possible to evaluate posture and detect alterations such as scoliosis,” he said.

Also in occupational health and safety, the solution can be used in ergonomics projects, where the environment should fit the worker rather than the other way around. “We want to help employers monitor individual body measurements where this is pertinent to the type of work. The solution can provide tools that make a supervisor’s job much easier,” he said

Micro and small enterprises are the main target market. A basic subscription is expected to cost between BRL 30 and BRL 50 (about USD 5.00-9.00). “We want to reach as many people as possible,” he said, adding that the goal is 5% of the potential market within five years.

In parallel with development of B-YOU, Inovia has won a government contract to develop a solution with computer vision. “It will be a system that analyzes X-rays of goods in containers for the Brazilian Customs and Federal Revenue Service. We’ll be paid BRL 4.2 million over a two-year period,” he said.

*Image by PublicDomainPictures on Pixabay