Data Science and Medical Technology: Automotive says hello!

Data science is a topic that has been on everyone's lips for a long time. Collecting data, analyzing it and ultimately using it for innovations is common practice in the market today - especially in the automotive industry. Vehicles are basically smartphones on wheels, so full of technology that a car mechanic from the 1980s would be speechless in the face of so much innovation - and his competence would be severely tested. Too fast the development, too fast the progress. The amount of time spent in cars today, saturated with data, can even be converted into hard cash: Researchers attribute up to five billion euros in research and sales volume to the minute that drivers spend in their cars and during which they use the car's services.


Data Science Professional


Ca. 7 min

Sharing is caring!

Data Science on the Industry 4.0 market

In short, the market is huge, and Data Science, with all its sub-fields, including Artificial Intelligence, Pattern Recognition, NLP, and Machine Learning, faces a rosy, because profitable, future: More possibilities, more scale, more complexity, and: More players. Pure data science companies are entering the automotive market because their experience in data science can be easily applied to the automotive sector. And that’s not all: automotive manufacturers and suppliers often lack experience and infrastructure, especially for purely cloud-based applications. Interesting side note: Telecommunications companies in particular could be among the serious players in the industry, as their wealth of experience in data science and cloud makes them leading global players.

Data Science: How Cognizant Mobility is thinking outside the box with automotive experience

Now we already talked about how new actors enter the stage, in line with Shakespeare’s motto that all the world is a (lifting) stage on which all people are mere players. Proactivity pays off more than ever, which is also the Cognizant Mobility has recognized this early on and has once again proven it, not least through the merger with Cognizant: Companies from the industry can no longer just follow prefabricated paths, focus on their core competencies and see themselves as pure suppliers. Rather, modern automotive players must become IT companies, cloudify, value data science – indeed, grasp the future. What sounds like a platitude at first is based on practical project experience.

Data Science in medical technology can work with experience from the automotive industry.

Thanks to Data Science: Automotive Conquers Medicine

Data science solutions developed in the automotive sector are ideally suited for recognizing and evaluating pattern-based events in other areas as well. So, as is so often the case, there is not just one path: It is not just data science and telecommunications companies that are pushing into the automotive sector – the forward-looking automotive stakeholders are also breaking into other areas, such as medical technology. One solution that Cognizant Mobility, as part of the Cogizant Group, is working on in this very area turned out to be as follows:

A leading medical player needed data on patients, basically times of a simple nature: inhale, exhale, lift a leg, an arm. The idea was to detect events in signals that would allow physiological measurements of a logical process. Of course, this also requires corresponding data sets, although these are not always easy to obtain: After all, the normal situation is that the patient does not move. Thus, the exception must be recognized, for which comprehensibly only small amounts of data are available.

The desired situation faced a problem that is quite relevant to medical technology, but which the automotive industry solved a long time ago, namely the detection of states. The basic principle of evaluating complex error patterns in BUS signals is comparable to the evaluation of complex movement sequences in patients. In individual cases, this can be done using neural networks or more classical methods such as decision trees. Crucial to robust signal processing is a time-invariant method that can detect the event regardless of when it occurs. One solution to this is the Bag-Of-SFA-Symbols (BOSS) method, which translates a signal time series into a language processing problem in an artificial language. For this purpose, the words generated by BOSS are further processed into a bag of words, i.e. without sequence or chronological order. The “winning algorithm” that was best suited for the final solution is the One-Nearest-Neighbor, which classifies data points based on the most similar data point previously classified. Thus, one looks at the most similar point about which one has already gained knowledge, and applies that knowledge to the next similar point, allowing for ever more advanced validation.

The Cognizant Mobility solution not only led to better results and can be applied in a generalized manner to signal sections of a different length or cut, which also absolutely satisfied the leading customer mentioned at the beginning and enabled him to immediately deploy the solution developed by Cognizant Mobility. Moreover, the signals no longer have to be recognized and entered manually; this is now done by the Cognizant Mobility solution, and the patient, customer and manufacturer all benefit: a successful growth from the automotive framework, with a precision landing in the medical research sector.

Data science also enriches healthcare research thanks to learning processes from the automotive sector

The future of data science outside the automotive industry

So while Data Science continues to evolve and new players enter the market, established players in the market must also not fail to keep innovating themselves, leaving well-trodden paths and entering new markets beyond the cozy comfort zone. Because the future is bright: A much finer-grained classification of data is on its way, multiclass cases are more than just dreams of the future: Logical craft instead of tepid vision. There will always be more data, more valuable, more difficult, more exciting. And it is obvious that automotive manufacturers are indispensable companions here due to their extensive data science experience: More and more complex problems require more and more complex models.

Which we develop for you. Feel free to contact us, right here using the contact form, or in the old-fashioned, but also incredibly human way: give us a call and we’ll talk – looking ahead.