This transversal exploratory study was conducted from August to December 2022 at the Universidade Federal de Ciências da Saúde de Porto Alegre to evaluate the feasibility and initial outcomes of the wearable sensor prototype. The transversal design was chosen to provide a snapshot of the prototype’s performance within a specific time frame. The participants were recruited according to predefined inclusion criteria, which typically included healthy individuals aged between 20 and 30 years without physical exercise contraindications. Exclusion criteria, if any, were also clearly delineated to ensure the safety and integrity of the study participants (Multimedia Appendix 1).

The study included healthy adults aged between 20 and 30 years without medical contraindications for physical exercise. Exclusion criteria comprised individuals with visual impairments, those with chronic conditions such as musculoskeletal or neurological diseases that could impede participation in the exercise protocol, and those who could not read or write. Since the study is an exploratory evaluation of a wearable sensor prototype, participants were recruited through convenience sampling using social media platforms, specifically WhatsApp (Meta). Invitations were distributed through various social groups and individual contacts on the platform, targeting individuals who met the predefined inclusion criteria. Before accepting the invitation, potential participants were informed about the study’s objectives, procedures, and eligibility requirements. Only those who responded positively and met the inclusion criteria were enrolled in the study, totaling 20 participants. This sample size was chosen to provide sufficient data to assess the feasibility and initial outcomes of the prototype within the constraints of the study’s scope and resources.

The prototype operates based on HR measurement using PPG. The peak-to-peak interval of the PPG signal is used to detect the HR. However, motion artifacts can contaminate the PPG signal during physical activity, interfering with HR estimation. These artifacts primarily result from ambient movement. Some filters were applied to address this issue.

The prototype uses PPG technology and incorporates a traditional PPG HR extraction algorithm based on the discrete Fourier transform [25]. PPG is a method that measures changes in light absorption corresponding to arterial blood volume fluctuations during systole using optical measurements [26,27]. The peak-to-peak interval of the PPG signal is used to detect the HR. Similar to the technology used in pulse oximeters, the sensor relies on optical techniques to estimate HR [26,28]. It consists of a photodetector and a light source that illuminates the skin to detect variations in light caused by changes in skin blood flow [28]. However, motion artifacts can harm the PPG signal during physical activity, interfering with HR measurement. These artifacts primarily result from ambient movement. Some filters were applied to address this issue. The collected data were stored in the cloud. Following prototype development, the LoRaWAN network was selected for transmitting sensor data over an exercise area of up to 10 km. After preprocessing, the sensor data are transmitted through LoRaWAN to a central concentrator, where they are encrypted and subsequently sent to a secure cloud-based database. Each user’s data history is stored in this cloud database, allowing for sensor validation, statistical analysis, and future integration with gamification interfaces. Data were continuously collected at a sample rate of 500 Hz and securely stored in the cloud, capturing detailed information with each heartbeat. Figure 1 illustrates the schematic diagram of the wearable sensor prototype, including its circuit, and provides a photo of the physical prototype. The prototype weighs approximately 150 grams and measures 5 × 7 centimeters.

Following the development of the wearable sensor prototype for HR monitoring, a validation test was conducted using the Incremental Shuttle Walk Test (ISWT), a well-established and safe walking test for assessing functional capacity [28,29]. The participants underwent 2 ISWTs, with the second test administered 7 to 14 days after the initial examination. Figure 2 presents the flowchart of inclusion and systematic procedures.

The ISWT involves graded bidirectional movement along a 10-meter corridor in response to audio cues. The standard test consists of 12 one-minute stages, starting at 0.5 m/s and increasing by 0.17 m/s each minute [29,30]. This study modified the protocol to include 3 additional stages, potentially allowing healthy participants to reach maximum exertion [31]. The test concludes when participants signal their inability to continue or fail to maintain the pace [30]. During the ISWT, participants wore the wearable sensor prototype on their nondominant wrist and the Polar H10 chest strap (Polar Electro Oy), the HR reference [17,18]. HR data were collected at three points: (1) at three minutes pretest, (2) during the test, and (3) at three minutes posttest. The average HR collected during each period was used in the analyses.

The quantitative outcomes were depicted as means and SD. The normality of the data was assessed using the Shapiro-Wilk test.

The 2-tailed t test was used to compare HR measurements from the Polar H10 and the wearable sensor prototype. Agreement between these measurements was determined using the intraclass correlation coefficient (ICC_3,1), with interpretations categorized as (1) <0.5=poor; (2) 0.5-0.75=moderate; (3) 0.75-0.9=good; and (4) >0.9=excellent [32]. In addition, the Lin concordance correlation coefficient was used to assess agreement between methods, with interpretations of (1) <0.9=poor; (2) 0.9-0.95=moderate; (3) 0.95-0.99=substantial; (4) >0.99=almost perfect [33].

The mean absolute percentage error (MAPE) was calculated to evaluate reliability or validity, with MAPE values ≤5% indicating high reliability or validity [34]. To calculate the effect size of agreement, Cohen d was used, with interpretations of (1) ≥1=very large; (2) 0.8=large; (3) 0.5=medium; (4) 0.2=small [35]. Bland-Altman plots were used to display agreement upper and lower limits and bias (mean difference), following the approach described by Bland and Altman [36]. A significance level of .05 was set for all tests. All analyses were conducted using IBM SPSS Statistics for Windows (version 27.0; IBM Corp.).

The study conformed to the resolution 466/2012 of the Brazilian National Health Council. It was approved by the local ethical committee for Research on Human Beings at the Universidade Federal de Ciências da Saúde de Porto Alegre (approval 54492221.80000.5345). Before participation, all individuals provided informed consent by signing the Informed Consent Form. No financial incentives were given to participants in the research.

Our analysis revealed a significant agreement between the HR measurements taken by our developed wearable sensor prototype and the Polar H10 strap. The wearable sensor is a valid instrument for HR monitoring during rest, exercise, and recovery periods. However, the prototype does not accurately measure maximum HR.

Low-cost sensors (priced below US $100) are pivotal in clinical practice and rehabilitation. They aid in physiological measurements for diagnosis and evidence-based practices and are vital for gamified apps promoting physical activity and rehabilitation [37]. Our findings advance the incorporation of low-cost sensor feedback into gamified apps.

We opted for LoRaWAN technology due to its cost-effectiveness and expansive range. Furthermore, the ease of its integration with smartphones made it a compelling choice [38].

Monitoring HR is pivotal for accurate exercise intensity determination and prescription. It is a proven method for athletes, healthy individuals, and patients with cardiovascular conditions [10,11,39-42]. Our results corroborate that our wearable sensor is reliable, with excellent ICCs to rest HR, mean test HR, and recovery. According to the Lin concordance correlation coefficient, the agreement ranges from substantial to moderate in the same variables [33].

Our error percentages are in alignment with the established literature. For instance, Fuller et al [13] showcased that 56.5% of wearable HR measurements had an error within 3%. Our prototype displayed consistent accuracy with MAPE values lower than 3%, aligning with previous research stipulating such thresholds for reliability and validity [13].

We observed the most significant MAPE during resting periods. However, as the literature suggests, pulse sensors often exhibit reduced MAPE during escalated speeds [15,42]. This behavior matches our findings during the ISWT.

The agreement was within the Bland-Altman plots’ upper and lower limits [36], but some tests were outside this range. This outside may be due to the difference in the technology used by the Polar H10 and the prototype that used the PPG. In this case, PPG technology has a limitation of susceptibility to motion artifacts caused by hand movement and differences in photosensitivity between individuals, which may limit data precision [43,44]. In the Bland-Altman plots [36], the agreement must be evaluated from a clinical point of view. In this context, the error HR is tolerable and does not limit the use of the wearable sensor prototype in exercise.

Our study has some potential limitations: (1) it was conducted in a controlled laboratory environment, which may not fully capture real-world conditions; (2) participants were healthy and belonged to a younger age group, limiting the generalizability of our findings to broader populations; (3) temperature conditions were controlled between 20 and 24°C, and we did not explore how the device performs under varying environmental conditions, which could impact its reliability; and (4) we did not compare the wearable sensor with others across different temperatures. These factors limit the applicability of our findings to diverse settings, age groups, and health statuses. Future studies will address these limitations by testing the wearable sensor in uncontrolled environments and across a broader range of populations and conditions, including older adults and individuals with varying health statuses. In addition, further investigation is needed to assess how different environmental factors, such as temperature, affect sensor performance.

One of the key strengths of our study is the creation of a low-cost wearable sensor that can be integrated into a gamified app. This approach not only makes the technology accessible but also encourages user engagement through interactive features. The wearable’s accurate HR measurement ensures safe exercise intensity recommendations, making it a valuable tool for personalized fitness monitoring.

Our device is precisely engineered to seamlessly integrate with gamified apps, enhancing user experience through real-time HR biofeedback. The traditional devices are often limited to brand-specific apps, our prototype leverages an open database architecture that allows for flexibility and interoperability with various platforms. It transmits HR signals through an antenna and stores them in a cloud-based database, enabling real-time access and processing of HR data. It is essential to effectively implement gamification strategies. Key features that make our device particularly suitable for this integration include the following. First, real-time HR biofeedback: the ability to transmit HR data in real time is crucial for interactive gamified experiences, allowing users to receive instant feedback on their performance and adjust their activity accordingly. Second, open database architecture: the flexibility of our open architecture enables compatibility with various fitness and gamification platforms, supporting the customization of exercise routines, challenges, and user profiles based on real-time data. This flexibility allows third-party developers to easily integrate the device into their systems. Third, cloud-based scalability: the cloud storage system provides secure and scalable data management and supports advanced analytics. This enables integration with machine learning algorithms that dynamically adjust game elements (eg, difficulty levels, rewards) based on user performance and trends. Fourth, personalization and adaptability: our device offers personalized feedback by analyzing the user’s HR data, making it highly adaptive to individual needs. This enhances engagement by offering customized rewards, progress tracking, and social interaction—proven elements of effective gamification. These features empower users by fostering a more interactive and responsive fitness ecosystem. The combination of real-time data processing, personalized feedback, and cloud-based analytics uniquely equip the device to enhance motivation and adherence to health goals through gamification.

This study is a pilot, and we plan to conduct further research to refine both the wearable sensor and the algorithm used for HR measurement, specifically using the discrete Fourier transform to improve accuracy, particularly when measuring maximum HR. The prototype does not consistently measure maximum HR with optimal precision. This may be due to 2 primary factors: further refinement of the PPG algorithm and motion artifacts during high-intensity exercise. To address these limitations, future improvements will focus on enhancing the PPG algorithm to filter out noise better and more accurately track HR at higher intensities. In addition, we plan to implement techniques to mitigate motion artifacts, such as using more advanced filtering methods or improving the sensor’s attachment to the body to reduce movement interference. These enhancements are expected to improve the sensor’s overall performance and accuracy in measuring maximum HR.

After refining the algorithm, we also plan to reduce the device’s weight and size, addressing other ergonomic issues. We will use 3D printing technology to help refine the design and reduce the size. This iteration will enhance ergonomic comfort and make the device more convenient and practical for users during physical exercise.

The production cost of the wearable sensor prototype was approximately US $38.50, excluding the antenna. At this stage, our objective is not to conduct a cost-effectiveness study, as the wearable sensor prototype is still under development. Therefore, we cannot present a final price or provide a detailed cost comparison with other commercially available devices. In future research, we plan to perform a cost-effectiveness study and compare our prototype’s costs and capabilities with similar devices.

In conclusion, our wearable sensor prototype effectively measures HR, drawing parallels with the Polar H10 sensor for rest HR during testing and recovery in the laboratory environment. Future work will involve integrating this wearable sensor prototype into gamified apps based on the validation performed in this work. This integration is expected to enhance adherence to regular exercise and ensure accurate intensity prescription, thereby maximizing the potential benefits for users.