The Key Takeaway
Minnetronix’s Clinical Trial study is an example of how Informative Analysis techniques can be integrated into a non-algorithm-based medical device to support improved functionality, proof of clinical efficacy, and future use cases.
In the case of Clinical Trial, our robust data science selection and development processes resulted in:
- Added value to the program by extracting additional statistically significant insights from the clinical trial above-and-beyond the device’s requirements.
- Prevented schedule and cost overruns by determining root causes of a key device development issue via regimented data collection and analysis.
- Maximized long-term value by uncovering and quantifying insights that guided technical and financial decisions for the device’s roadmap.
Note: to respect non-disclosures, we’ll refer to this study as ‘Clinical Trial’ and the device being studied as ‘Device’.
Unlocking Value with Data Science
Data is a crux of any medical device’s journey but can often be a barrier to success. For example, early-stage research and development may yield upwards of terabytes of data that can provide insights into the efficacy of a device. By failing to effectively analyze this data for meaningful insights that could influence the direction of the program, medical device companies risk leaving significant future value on the table. Involving data scientists early in the device’s development program can help teams effectively plan for the future and mitigate the cost of change. And following a rigorous data science process can help teams unlock the maximum value from their costly input data.
Minnetronix offers full-scale data science capabilities as a part of our core services. We are a single partner that excels at the junction of medical device and data science, meaning that our clients don’t need to manage the data science themselves or seek additional support from an outside vendor. When partnering with Minnetronix, our data science processes are built into the core medical device development program and handled by our team of data scientists. Instead of running siloed workstreams that merge at the end. This integrated approach ensures better overall device performance and higher lifetime ROI for developers.
A good example of how our data scientists can deploy Informative Analysis techniques to create value for a medical device program is highlighted by this Clinical Trial.
“Our data science processes are built into the core medical device development program and handled by our team of data scientists.”
Answering Questions with Clinical Trial
The value of our data analysis capabilities are highlighted by this Clinical Trial Minnetronix supported for a medical device, hereby referred to as Device. This answers a critical question for medical device developers: “How might data science capabilities be leveraged to solve problems in my project outside of algorithm development?”
Minnetronix generated data in the initial research and development of the Device. This data, when interpreted, provided key insights into device efficacy and helped facilitate commercialization.
Clinical Trial was a feasibility study that enrolled dozens of patients, not hundreds. However, Minnetronix recognized the richness of the data gathered for Clinical Trial to explore and validate the Device’s potential. We maximized its utility via our informative analysis approach performed by our data scientists and following our proven process. Our work surrounding the Device also demonstrated our ability to address client needs and provide commercialization support.
“Our work surrounding the Device also demonstrated our ability to address client needs and provide commercialization support.”
Selecting the Data Science Approach
There are many different data science techniques – different ways of analyzing data to extract value from a data set, or different methodologies to algorithm development. Therefore, one of the most critical parts of any data science effort is selecting the technique that is the best fit for the application.
Minnetronix has developed a framework to aid in this selection process and provide medical device development teams with a simplified approach that minimizes future risk. Our framework considers four general approaches to data science, each of which provides benefits and drawbacks according to six competing characteristics. The correct choice helps us minimize cost of change and navigate trade-offs that will happen downstream in the development process.
After strategically analyzing the data and business needs for Clinical Trial, we determined that the informative analysis approach was optimal to maximize efficacy while managing the schedule and budget. By using our proven framework to aid in the selection process, we made choices that resulted in improved flexibility and the best minimization of future risk.
Assessing the Needs for Clinical Trial
We chose the proper approach based on six competing variables.
The amount of data needed to gain a thorough analysis helps determine which approach is best. While some devices need massive data sets, others will suffice with less. Approaches with fewer data requirements tend to lead to lower costs and faster computation speeds.
Clinical Trial’s data set was small because of the stage of the product (early human use, not yet cleared) and because access to the data was expensive. Therefore, our data scientists ensured we were maximizing the potential of this limited data set.
The data science work for Clinical Trial needed to be explainable to a variety of key stakeholders: internal leadership, company board of directors, clinical key opinion leaders, and editors at medical journals. Therefore, one of the logical next steps after data interpretation was the creation of a manuscript that highlights the capabilities of the Device. This requires the development of clear, concise data visualizations that communicate the findings in an easily interpretable fashion.
The speed of computation was not a major factor in decision making since proper off-the-shelf tools can produce statistical results in a negligible amount of time for small datasets such as this.
Cost and timeline certainty was prioritized in this Clinical Trial since analysis had to be complete in time for business milestones to be met. Lower-variance statistical testing was employed in place of higher-variance testing to mitigate cost and timeline risk inherent to small-sample analysis.
We didn’t employ more novel techniques here since explainable analysis for key stakeholders was a priority. Instead, standard statistical methods for the given data types were employed to make analysis understandable while remaining impactful.
Some data science approaches have a greater ability to grow following commercialization. AI-Neural Nets, for example, can be retrained to detect new diseases from the same imaging modality or can be retrained on a larger dataset to improve on previous performance.
For this Clinical Trial, this ability was not considered to be a requirement. There was no algorithm being developed; the data science work was focused on patient trajectory and outcomes as well as device performance.
Following Our Process
A standardized data science process is vital to success due to the high cost of input data for medical devices. As a result, teams need to ensure they make the most out of their scarce and/or expensive input data.
Minnetronix has over 25 years of medical device development and manufacturing experience, meaning we have relevant contextual knowledge that we apply to our programs. Not only do we have the streamlined process to gather the data, but we also have the knowledge of clinical applications to apply to it. When you work with Minnetronix, we will follow the data science path that maximizes the data you have to achieve your business goals.
In this instance there a few steps in our process that were most critical for the success of Clinical Trial.
Our team identified the need for data in a similar population as the Clinical Trial study. A comparable, retrospectively collected control arm (Control) provided an independent sample to compare Clinical Trial against. Findings from the data analysis and statistics would be presented and published in manuscripts that provided proof of value for Device.
To accomplish the data science portion of this project, we identified the following needs and individuals to fill these needs:
- Data Science Expertise (in-house)
- Clinical and Clinical Trial Expertise (in-house and partnership)
- Data Access (partnership)
The Minnetronix team gathered data from nearly 500 patients as part of the Control population. Here, our ongoing relationship with a hospital who participated in Clinical Trial benefitted our publication strategy. Over half of the subjects in Clinical Trial were enrolled at this site. Forming the Control from this site made a more representative comparator group than any other site and enhanced the credibility of the publication.
Clinical knowledge was applied to the datasets to identify inconsistencies, allowing data correction and a better comparison. Clinical knowledge and data science skills were applied to unify inclusion/exclusion criteria, outcome metrics, etc between Control and Clinical Trial. These steps aided scientific and clinical acceptance of comparison between two groups in a non-randomized control setting.
Before testing, we explored relationships between Control and the Clinical Trial data to draw insights from the information and determine the proper statistical testing methods. These insights informed aspects like the data distributions, sample sizes, and comparisons.
After building and evaluating the statistical models, we analyzed the results and designed graphics for easier understanding. We engaged physicians with these visuals, gathering feedback on what was most clinically meaningful for them and how we might improve the presentation of data to be most meaningful to fellow clinicians.
Several iterations of these visuals were made, leading to more understandable and meaningful figures to be used in the final publications.
We provided support for manuscripts on Control, Clinical Trial, and a comparison between the two to highlight the value of the product. These manuscripts are in the process of physician review and feedback to assure that they are as clinically meaningful as possible.
After commercial approval of the Device, these manuscripts will be submitted to their target journals for publication and will be used in support of the Device. Submission of these manuscripts is on track to be finished prior to commercial approval of the Device.
Working with Minnetronix
Minnetronix was successful in this project, and many others by stressing the value of a dedicated data science selection and development process. Our selection of the informative analysis category allowed for the quickest path to commercialization.
Considerations of the approach to data science selection led to the identification of a data gap. This gap would have otherwise led to a schedule delay for this medical device. Our process kept the team focused on supporting the commercialization of a viable product. This included a focus on identifying and filling gaps to create manuscripts that are meaningful to clinicians.
Our approach is collaborative. For example, our close relationships with clinicians in the Clinical Trial led to more, unplanned manuscripts in support of the Device. These manuscripts investigated future use-cases, market opportunities, and potential device optimizations. This additional value-add wouldn’t have been possible without the collaborative nature of our process.
Data capabilities and expertise can also provide value during the development process of a project. For instance, the verification of Device hit a snag when a pressure monitoring system wasn’t meeting the required accuracy. Many solutions by the engineering team were attempted but failed to fix the system. Minnetronix data scientists looked at the system with fresh eyes and used long test runs, logging, and data visualization to determine a confluence of multiple factors which were leading to the issue. This identification of the specific problems allowed for each of the issues to be addressed, reducing risk to the cost and timeline of the product.
This clinical study’s approach and Minnetronix’s overall data capabilities led to a smoother process for Clinical Trial. By offering these services to our clients, our team enables the commercialization of a product without complicating the overall production. The best part is that we offer these services as an integral part of our services–reducing your need to interpret the data yourself or to find an additional partner.
In the case of Clinical Trial, our robust data science selection and development processes resulted in
- Added value to the program by extracting additional statistically-significant insights from the clinical trial above-and-beyond the device’s requirements.
- Prevented schedule and cost overruns by determining root causes of a key device development issue via regimented data collection and analysis.
- Maximized long-term value by uncovering and quantifying insights that guided technical and financial decisions for the device’s roadmap.
With Minnetronix as a partner during development, you can leverage any and all data at your disposal by utilizing our proven process, knowledge of clinical applications, and in-house data scientists. Whatever the needs of your medical device program, Minnetronix data scientists add value and solve key development problems.
Interested in learning more? Click here to contact us!