Healthcare is one of the frontiers of artificial intelligence breakthroughs. In years to come, advanced learning algorithms will help researchers unravel more of life’s mysteries from strands of DNA. It will also help doctors scout for trouble in people’s genetic profiles and take a proactive approach to disease prevention.
But looking to the future should not obscure the reality of the progress we’re making today. Two use cases from DMI’s client portfolio illustrate these gains. In one, our AI development partner helped predict the outcomes of clinical trials for cancer drugs. In the other, we helped a health insurance giant improve customer service in its call center.
Prediction: Avoiding Adverse Outcomes in Clinical Trials
Establishing the safety and efficacy of new medications can take over a decade and cost billions of dollars. Speeding up this process while reducing costs drives value across a vast health care ecosystem of doctors, patients, providers and insurers.
A recent DMI project conducted with our AI partner revealed cause for optimism that AI can play a role in driving this kind of value. Our AI-powered project for clinical studies on cancer medications:
- Accurately predicted clinical trial outcomes at the clinical ARM/trial and molecular levels, reducing time frames, shrinking costs and improving treatment efficiency.
- Enabled researchers to predict therapeutic failures, helping patients avoid wasting time on ineffective treatments.
- Produced useful results with platform-agnostic methodologies that can work with any system.
The research project developed a predictive AI model based on a massive database of colorectal cancer cases. The project combined the colorectal research with the AI model and replicated the results in pancreatic cancer patients. Now the model can be applied to other tumor types.
These results point to the potential of predictive AI to help clinicians get more proactive and less reactive — avoiding the well-worn path of treating people only after they become sick. Preventive tactics like these may prove crucial to controlling the escalating costs of health care.
Conversational Interfaces: Improving Customer Service for Insurers
Another DMI client, a large health insurance provider, needed help transforming its call center. The company was sitting on seven years of unstructured data such as audio files from its government and commercial lines of business. They knew this data held a treasure trove of information about how people use the call center, so they put us to work on an AI-driven strategy to capitalize on it.
DMI’s AI solution:
- Used call log analytics to streamline call center operations. That produced higher customer satisfaction and loyalty while improving operational efficiency and boosting agent performance.
- Crafted a four- to six-week proof of concept (POC) that combined call log analytics with a root-cause analysis based on customer reactions, tone of voice, speech habits and hot topics that drive call-center traffic.
The project combined natural language processing (NLP) with text and sentiment analytics to determine what customers need most from a support call. When people call in with a simple problem, the conversational AI can implement a chatbot to direct them to a web page. If they have more complex problems, they get directed to expert agents who can help them figure things out.
Naturally, the most experienced and effective call center employees earn the most pay and represent major cost centers. Our system automatically rerouted minor calls away from them, giving them more time to deal with complex problems. That’s a win-win-win for customers, skilled employees and the company.
Voice and Prediction are the Future of AI
While these examples are specific to today’s health care challenges, they represent the path tomorrow’s AI will take.
Natural language processing and predictive AI use advanced pattern-matching algorithms that essentially teach themselves to seek success and avoid failure. Together, they will enable heavy industries to forecast commodity needs and prevent expensive production shutdowns. In the consumer sector, they’ll help marketers find buyers with laser precision.
Of course, AI is not a gold-paved road to utopia. A host of difficult security, privacy and governance challenges have to be ironed out. But the progress in health care suggests advanced algorithms can be a force for good for people, professions and the economy. It all depends our intelligence in putting these tools to their best use.
-Niraj Patel, SVP IoT