AI’s Potential to Transform Cars, Food and Healthcare
Many industry giants are running out of options with increased competition. Wall Street is pressuring such companies to outpace potential disruptors or, like Ford CEO Mark Fields, get booted. Disruptors also grow — and get huge valuations — quickly, making buying your way to innovation less attractive.
One alternative is to adopt the tools like advanced data analytics and artificial intelligence to create better efficiencies. Businesses that do so enjoy competitive advantages, significant cost savings — and most importantly, ensure they provide their customers with the most relevant product or service.
As we head into 2019, it’s hard to find an industry that has been untouched by the data revolution. Even segments known for the hands-on nature of its work, like construction, are being reimagined with 3D-printed buildings, augmented reality and robots. The three industries below stand somewhere between those most and least affected by digital transformation. Here’s a look at how the three — automotive, manufacturing/food production and health insurance — are embracing data analytics and artificial intelligence, often through IIoT innovations (Industrial Internet of Things) and the benefits each are realizing:
Autos: Different Warranties for Different Drivers
It should come as no surprise that not all drivers treat their vehicles the same. These days, many cars have sensors in their interiors. Such sensors are tracking everything about how drivers use their cars and that data can be used to better segment drivers. Everything from how many turns a driver takes to how often they brake gets logged in a database. It even measures the pressure on brakes to distinguish hard braking from soft braking.
Once you pull such data together and apply advanced analytics, you can see patterns of people who are doing more brakes and stops than average and who is driving hundreds of miles on the highway where they barely have to touch the car. Using such data can change an automaker’s focus from product segmentation to a more granular customer segmentation.
One thing you notice by taking this approach is that mileage is less of an accurate barometer for warranties than usage. The worst kind of usage for a car is when a vehicle is driven mostly in traffic while the best is when it’s mostly highway driving, with less stopping and starting. An automaker in that case can offer everyone the same warranty for the first year but then can offer a different warranty package in the next year based on usage. Though no automakers have done so yet, this type of warranty package (similar to how the car insurance industry uses in-car tracking devices) can save automakers a lot of money and reward drivers who are gentler on their cars. For example, one of our auto clients was able to reduce warranty costs by 35% using sensor data.
Autos: Big Data for Product Design
Similarly, sensor data could be used to inform new product design for auto manufacturers. While auto manufacturing is complex and there would be challenges around making real time changes in the production line, data inputs could be used for future product design. By taking all of the input from the sensor data, auto makers could focus on the features that consumers are actually using and save money on eliminating features they aren’t using. It could also be used to re-configure different systems. Examples of this include the entertainment and navigation systems, which has a lot of software powering it. Physical examples include things like cruise control or climate controls. If the data sensor data shows that drivers are more comfortable adjusting these things manually, then the manufacturer could evaluate how much they invest in these systems for future models. Data for product development is important as organizations that leverage customer behavioral insights outperform peers by 85% in sales growth and more than 25% in gross margin, according to research by McKinsey & Co.
Food Production: Minimizing Manufacturing Stops
In food production, the use of sensors in machinery aren’t new. Some date back to the 1980s or 1990s. But since data storage was historically so expensive, no one had been collecting such data in a central repository. Also, the factory engineer who traditionally collected such data weren’t data scientists, but rather manufacturing experts who were accustomed to monitoring pressure and temperature. Today, data storage costs have come way down with the ease of public cloud, and advanced data analytics to realize new efficiencies.
One way to make food production more efficient is to limit so-called manufacturing stops. That name is a bit of a misnomer since the line never actually comes to a complete stop. Instead, there are an endless number of instances where equipment along the production line stops for a short period of time due to obstructed product flow or something easily resolvable that doesn’t require maintenance personnel to get involved.
For example, think of the production and packaging line for a product like potato chips or soft drinks. If the there’s a stall in the line where the chips get bagged or the soft drink goes into the cans or bottles, the entire process slows down for what may be an unperceivable five seconds as the glitch gets fixed and the next batch of chips can come come through the line, or the soft drink start flowing again. These series of small delays in food production can add up.
According to a 2017 OEE Benchmark study via Epicor and Sage Clarity: “Minor stops are a chronic problem in the food and beverage industry, up to 60% of downtime. Minor stops are difficult to capture manually and require automatic data collection. Otherwise, companies are ignoring 60% of production problems.”
By analyzing all of the sensor data that is being generated continuously in real time along the production line, and using artificial intelligence to organize and analyze for efficiency and optimization, food production plants can create incremental production improvements every month and even single digit percentage increases can translate into millions of dollars over time for any individual manufacturing plant.
Healthcare: Personalized Medical Devices
AI and advanced analytics is rapidly transforming the healthcare industry from leveraging data for efficient diagnosis to pattern recognition for treatments to resolving logistical challenges. In the manufacturing realm, data analytics holds major potential for advancing the way medical devices are designed and produced for individuals. With flexible manufacturing processes like 3D printing, if you have the right data available that can be analyzed (even if it’s aggregated and anonymized), the industry can get closer to manufacturing devices – from custom braces in dentistry to medical devices that are implanted into patients. Instead of a limited set of models or devices, patients will have increased choice for devices that are customized for their specific needs.
This isn’t just theory either. According to a Medical Xpress article, “researchers at the Wyss Institute for Biologically Inspired Engineering at Harvard University have created a novel 3D printing workflow that allows cardiologists to evaluate how different valve sizes will interact with each patient's unique anatomy, before the medical procedure is actually performed. This protocol uses CT scan data to produce physical models of individual patients’ aortic valves, in addition to a "sizer" device to determine the perfect replacement valve size.”
Starting Small
The opportunities for advanced analytics to help businesses navigate through digital transformation are endless. The above examples demonstrate just three areas where advanced analytics and artificial intelligence have massive potential to transform the way various industries operate. It often turns out that there are lots of areas for businesses to save money, but when it comes to implementing AI solutions, it’s a good practice to focus on one area, see the results and then move to the next.
Tackling AI initiatives is a difficult proposition for many industries. It requires resources, people and an overall change to mindset with buy in from leadership. The first step is to start small and pick one area to start where you want to make a change. Then, if you don’t have the in-house capabilities and resources, you need to choose a partner that has the ability and expertise in both AI and domain expertise in your specific industry to consult and lay out a plan. Once you’ve established several examples and projections on the benefits that deep data analysis can have on your business, it becomes much easier to get full management buy in for implementation.
