This extract is part three of a five-part series that breaks down the layers of the Data Strategy Framework for Connected Medical Devices, which we introduced in our recent whitepaper. Featuring insights from industry leaders at Cochlear, Dexcom, Medtronic, Omron, Vocxi Health, and S3 Connected Health, the paper offers a practical framework for leveraging data to stay competitive and succeed in a data-driven healthcare landscape.
This layer serves as the engine that converts raw data into actionable intelligence. It is particularly critical for medical device vendors focused on smart connected patient devices. It processes the vast amounts of data generated by these devices and transforms it into insights that drive improvements in product performance, operational efficiency, and clinical outcomes.
This layer is where raw data from devices and other sources is analyzed using various analytical techniques, including data science, machine learning, and AI. The goal is to uncover patterns, trends, and insights that can guide business decisions, improve patient outcomes, and enhance operational efficiency.
Additionally, this layer contextualizes data by correlating it with other sources, providing a fuller view of the device, the user, and all involved stakeholders. This context is critical for driving meaningful actions.
Key functions of the data analytics and insights layer
Descriptive analytics: What happened?
Descriptive analytics provides an understanding of historical data, helping to identify trends and usage patterns over time. For medical devices, this can reveal how devices have impacted patient outcomes or how they’ve been used in real-world settings.
Example: A report on the usage patterns of wearable devices that monitors trends in heart rate, activity, and other vital signs over a specified period.
Diagnostic analytics: Why did it happen?
Diagnostic analytics delves deeper into the data to determine the causes behind specific patterns or events. For medical device vendors, this type of analysis can identify the root causes of device malfunctions or adverse patient outcomes.
Example: Investigating why an implanted device may have failed earlier than expected, potentially leading to a recall or design improvement.
Predictive analytics: What will happen?
Predictive analytics uses statistical models and machine learning to forecast future outcomes based on historical data. This is crucial for anticipating patient needs or device failures.
Example: Predicting the likelihood of a patient experiencing a heart issue based on data from a wearable ECG monitor or forecasting when a device may require maintenance or replacement based on usage patterns.
Prescriptive analytics: What should be done?
Prescriptive analytics suggests actions to optimize outcomes, often relying on AI-driven decision support systems. This type of analysis can guide treatment decisions or recommend preventive actions for medical device vendors.
Example: Recommending adjustments in insulin dosage for patients using a continuous glucose monitor or suggesting device recalibrations to improve patient outcomes.
Real-time analytics:
Real-time analytics processes data instantaneously, providing immediate feedback and enabling timely interventions. This is especially vital for medical devices that monitor critical health parameters in real time
Example: Continuous monitoring of an implanted defibrillator that detects irregular heart rhythms and triggers corrective actions or alerts in real time.
Context is key: Making data meaningful
In healthcare, context is critical to making data meaningful and actionable. While companies can gather more data than ever, the challenge lies in interpreting that data in ways that align with organizational goals and patient outcomes. Data must be organized and structured in a way that makes it useful. Even the most advanced data collection systems can become overwhelming and ineffective without a clear strategy to contextualize information. A good data strategy, therefore, is not just about collecting information but also about understanding its relevance and utility.
“The relevance of data and decision-making depends heavily on the specific health condition being addressed. No matter how seemingly irrelevant, every data point can contribute to a more accurate understanding of human behavior. This process of gathering and analyzing contextual data is time-consuming but crucial. Behavior is unpredictable, which is why continuous tweaking and reassessment are necessary if we want to drive adoption of our devices and patient adherence to new therapies.”
Harsimran Singh, Director of Behavioral & Translational Data Science, Dexcom
Paul Stevens, Director of Digital Health, Omron Healthcare Global, reinforces this point with an example from Omron's journey. He explains how their first cloud-based service, Omron Connect, resulted in a massive database of information that initially lacked context: "We had tens of millions of connected devices on the market collecting data, but what we realized over time was that none of that data was contextualized." By shifting focus, Omron was able to transform raw data into valuable insights, such as showing users how their medication affects their blood pressure. Integrating external data sources like Apple Health and Google Health (with patient consent) further enriched insights, improving patient outcomes and services.
The field of behavioral science brings extra nuance to contextualizing data, as experts understand that every data point can contribute to a more accurate understanding of human behavior. Context is not static but changes with behavior, so data must be constantly analyzed to make informed decisions. It’s crucial for healthcare companies to not only gather data but also continually refine its interpretation to drive meaningful health improvements and enhance patient adherence.
"Gathering and creating data is good, but we must pull that together. We need to organize it, structure it, and make it useful. The promise of digital health is that I can look at myself 24/7, 365 days a year, from several angles, but that doesn't do me any good if I have to open 5 apps to look at all of those different things. If you can’t turn data into actionable insights, then what's the point? Companies need to focus on making data meaningful, so it leads to actions that make a difference.”
Bill Betten, Director of Medtech Solutions, S3 Connected Health
Challenges and considerations
Data volume and complexity
Medical devices, particularly smart connected patient devices, generate vast amounts of data. Managing this volume and ensuring that it can be processed efficiently requires robust infrastructure and advanced algorithms.
Data quality and integrity
High-quality, clean data is essential for accurate insights. Inconsistent or incomplete data can lead to incorrect conclusions, potentially harming patient safety or creating poor business decisions.
Interpretable and actionable insights
The analytics must produce results that are understandable and actionable for non-technical stakeholders. For medical device vendors, this is crucial in ensuring that clinicians, business leaders, and operational teams can use the insights to make informed decisions.
AI bias and fairness
Machine learning models can sometimes perpetuate biases present in the data. Ensuring fairness in healthcare decisions is critical to prevent biased treatment recommendations or diagnoses.
Check out our recent whitepaper for more information on building an effective data strategy framework for connected medical devices. Featuring insights from industry leaders at Cochlear, Dexcom, Medtronic, Omron, Vocxi Health, and S3 Connected Health, the paper offers a practical framework for leveraging data to stay competitive and succeed in a data-driven healthcare landscape. You can read other extracts in this series to understand each layer better:
