We've updated our Privacy Policy to make it clearer how we use your personal data.

We use cookies to provide you with a better experience. You can read our Cookie Policy here.

Rectangle Image
Industry Insight

Five Steps To Create Wearables That Can Diagnose Disease Onset

Rectangle Image
Industry Insight

Five Steps To Create Wearables That Can Diagnose Disease Onset

Fitbits, Apple Watches, smartphone apps and other products have become ubiquitous in consumer health tracking. In fact, the Consumer Technology Association (CTA) projected that the total revenue of health and fitness tech will reach $13 billion in 2021, a 12% growth over last year, with smartwatches seeing an 8% growth in unit shipments. As these technologies quantify and capture the behaviors and physiology of consumers, they are generating an abundance of wearable and voiceprint data. These data contain a wealth of information that can enable early detection of diseases like diabetes, asthma and heart conditions, as well as rarer conditions like Parkinson’s and amyotrophic lateral sclerosis (ALS).

A wearable and voiceprint data analysis strategy will allow researchers and health professionals to capture data, identify trends of interest and create algorithms and tools for discovery. Taking the appropriate steps to implement this strategy will be essential to their success.

Creating a wearable and voiceprint data analysis strategy

Though consumer health apps and devices can generate valuable proof points for clinical trial research that are indicative of real-world patient behavior, developers must go beyond simply providing these devices to patients. A wearable and voiceprint data strategy entails monitoring data flow to deftly decide when to utilize and analyze the data from wearables. By following these five key steps, researchers can create an effective wearable and voiceprint data strategy and make the right decisions about when to leverage these data.

1. Study a specific patient population for extended perios

To begin, researchers will need to define a baseline patient population from which to collect data for an extended period of time. A specific criterion should be created to select the proper patient population for the research. This will assist in providing developers with the data needed to legitimize common health trends.

2. Analyze the data with artificial intelligence and machine learning (AI/ML) technology

The combination of the growth of wearables and the fact that these devices are continuously generating data means there is an ever-growing wealth of evidence to handle and analyze.  Employing AI/ML technology will aid researchers in analyzing it at scale to discover the parallel health outcomes, trends and criteria related to imminent health events.

3. Identify trends through ML algorithms

Developers should train ML algorithms to proactively assist in seeking out potential health events using baseline data. When training ML algorithms, developers should consider the possibility of false positives and negatives and find supporting data to weaken the possibility of those errors, and/or to alert users of the risks.

4. Generate a real-time response system

Once common trends are found and verified through the ML algorithms, implementing a real-time response system will ensure the parties for whom these insights are most relevant are notified, whether that be patients or their physician or clinical research investigators. This alert system will aid in taking the next step to create preventive actions that can reduce negative outcomes, and lower related healthcare costs.

5. Determine when to use wearables

Wearable and voiceprint data can be used on a case-by-case basis depending on the goals of the researchers. For example, wearables and voiceprints in human trials can provide real-time data about patient conditions to provide evidence on the impact and success of a treatment, as well as early warnings of adverse events. Alternatively, post-market use can
serve as a comparator to demonstrate real-world effectiveness.

Data recorded from wearables also need to be carefully analyzed for developers to accurately train algorithms for the specific case being studied. Although
studies have shown the clinical impact and effectiveness of wearables for a variety of patient health conditions, whether wearables data can be generalized across other conditions remains unclear, so it is important to assess if using data from wearables is right for a study.

Preventing adverse events

After completing the initial analysis steps of the wearable and voiceprint data analysis strategy, investigators can use data and trained algorithms to further their comprehension of patients’ health in clinical studies. Data findings will be critical in predicting potential adverse events and accelerating knowledge of whether a drug or device is working in sub-populations in a study to help prove the impact of a drug or device to regulators. Ultimately, the data will impact treatment patterns, name early warnings of adverse events and help investigators find sub-populations who are responding positively to the treatment.

The future of health outcomes through wearables

The possibilities of wearables and consumer device data to investigators seem endless. However, for researchers to rest assured that they are logging the right trends and reducing the risks of false positive and negatives, the proper data strategy to encapsulate the meaning of data is crucial. When properly analyzed and utilized, these data will ultimately improve safety and health outcomes, and moreover, serve the greater pursuit of patient centricity in medicine.

Kal Chaudhuri is principal, AI/ML applications, R&D strategy practice at IQVIA Real World Solutions