Artificial intelligence (AI) is beginning to radically transform every industry, and these opportunities are immense for healthcare. Clinical applications have been at the forefront of adoption: AI to predict and prevent readmissions, manage chronic diseases, and drive clinical decision support tools.
See for yourself how ReferralMD can improve your patient referral process with patient engagement tools that reduce no-shows.
Now, top companies are beginning to develop AI and machine learning tools to improve operations at hospitals, clinics, and practices. Innovative leaders can implement these next-generation tools to capture more value today and future-proof their business.
What is AI and Machine Learning?
Like many advanced and rapidly evolving technologies, artificial intelligence has many definitions. According to Merriam-Webster, artificial intelligence is the power of a machine to copy intelligent human behavior. In healthcare, AI refers to human-level or greater intelligence applied to very specific tasks, such as knowledge management, speech recognition, natural language processing, computer vision, and more.
One of the most important subareas of AI is machine learning, a major field within computer science, which aims to build computer systems that automatically improve with experience. When machine learning is applied to a task, it can learn how to perform the task and improve its accuracy, output, or speed over time.
Why is it important?
Imagine training an employee who never forgot, never retired, never took a break, and continuously learned on the job. Specific applications of AI and machine learning make this extraordinary productivity a reality across a wide variety of operational tasks.
How can we use AI today to improve healthcare?
Executives and physicians ask: How have hospitals begun to adopt these technologies? What are the benefits? And how can my hospital leverage AI? The following applications for operations explain how hospitals can use AI today to reduce costs, improve patient outcomes, and transform healthcare.
AI tools to improve patient access
A key driver of success for providers is the ability of patients to see their caregivers in a timely manner. Patient access is a major priority for leading hospitals and practices across the country. If your patients cannot get an appointment with primary care, specialists, or surgeons promptly, they will go to other providers. Referred patients who cannot be seen quickly will lead to reduced referral rates, decreased revenues, and lower patient satisfaction scores.
By employing new AI-based technologies, your organization can significantly improve patient access. These artificially intelligent systems utilize predictive analytics to help your teams make better decisions.
1. Smart scheduling for outpatient appointments
Scheduling is laborious, often mismatched to true needs, and seen as a chore for many care teams. Today, schedules are made around set time blocks, but the needs of each patient rarely fit into pre-defined 15 or 30-minute time slots. More importantly, scheduling tools cannot help predict no-shows and cancellations nor guide appropriate follow-up actions.
Reports show that a single no-show can cost your practice $200 and the total cost of no-shows for the US Healthcare system is a $150B problem. It is important to understand how long each patient will need, determine who will be a no-show or cancellation, and improve the patient flow through your office. Your schedule is the intake to your entire practice, and when investments are made in better access, your patients get scheduled faster and spend less time waiting in the office.
By leveraging machine learning, you can understand patient demands, plan for the unexpected, and improve patient access. These predictive analytics are multifaceted and offer two-fold benefits:
- Predict no-shows and cancellations: using past patient data, patient demographics, chief complaints, location, and environmental factors, algorithms can predict who will show up late, who will cancel, and who will no-show. These aberrations to your calendar mean lost revenue, patients missing critical care, and reduced access for new patients.Adopting predictive analytics for your schedule improves patient throughput and alerts your staff to trouble. By flagging likely no-shows, your team can proactively reach out to make sure a patient who needs care can come in. When necessary, it can also recommend double-bookings to keep your schedule optimized.
- Determine the appropriate appointment length: using the patient record, patient history, and physician habits, machine learning tools help your staff set the right appointment length. Some of the best schedulers may have a deeper understanding of their patient population and know the true length required of the appointment given the particular patient.With predictive analytics these factors are known and estimated for every patient accurately. By right-sizing your appointment times, you can see more patients, have a more predictable schedule, reduce wait times, and increase satisfaction.
2. Smart scheduling for the OR, GI, and Cath labs
The operating rooms, GI labs, and Cath labs are some of the most important service areas your hospital maintains. These are the profit-drivers of your hospital, but they can also be major cost sinks if not carefully managed. Stanford studies show the average cost of ORs can be as much as $60-80 or more per minute.
Procedures that run long can lead to costly overtime wages, dissatisfied staff, and may bump planned operations out of the schedule. Under-utilization of these rooms leaves revenue on the table and keeps patients and surgeons waiting. Unnecessary gaps in the schedule also lead to additional delays with staff taking breaks outside the operating theaters.
If your team better understands and predicts the needs of the OR, GI, and Cath lab, you can increase throughput, improve revenues, reduce costs, and improve physician and patient access.
Novel machine learning techniques can ingest all your operating room scheduling data to generate insights. By tracking time-stamped data for arrivals, preparation, and procedures by patient and physician, true operation and turn-around times can be predicted. This tracking will enable your schedulers to allocate the right amount of time for each procedure based on the individual patient and physician.
AI systems can suggest the ideal schedule and remove gap time from your hospital’s most valuable assets. Further, they can specifically optimize block time or help your team move off blocks entirely, while still prioritizing key value-drivers. A smart schedule, powered by AI, will drive efficiency in your OR and critical procedure rooms.
AI tools to support finance
Your practice or hospital’s financial well-being is paramount to keeping its doors open. It is time to arm your finance team with the same firepower that payers have been using to limit and control payments. The cat and mouse game between providers and payers has led to shrinking margins and difficulty getting the money that is due. AI is perfectly suited to help your finance team work through the myriad hoops and data requests payers burden on providers
3. Automated Pre-authorization
Pre-authorization can take an army of trained clerks to process clinical notes, answer questions for payers, track dozens of contracts, and not let anything slip through the cracks. According to studies by the American Medical Association and the Journal of the American Board of Family Medicine, today’s pre-authorization methods can take 20 hours of labor per week per physician and lead to a total annual cost of $600-700M in the US. Preauthorization is a major financial burden.
Many for-profit hospitals won’t schedule imaging or procedural appointments without pre-authorization, and delays in the process might turn away patients to other facilities. Some nonprofit hospitals that don’t wait for preauthorization put reimbursement directly at risk. Worse, after critical clinical documentation is captured by clerks for pre-authorizations, they are not retained for use in claims.
Fortunately, AI can help overhaul and help automate the entire process. Using advanced natural language processing, contextual understanding, and large training sets, AI technologies automatically identify and collect key clinical information from the patient record. These tools prepare summaries of information for the clerks to help them answer questions faster and more accurately. These same engines can be used to automate all but the hardest cases to quickly and efficiently get pre-authorizations.
Leveraging AI to help tackle the bulk of pre-authorizations enables your team to focus on the smaller pool of complex cases and frees it up for more important administrative tasks. Further, with an automated engine to track and collect data, your team will have better documentation for claims processing downstream and decrease the time it takes for you to get paid.
4. Predictive denial management
Denial management is a critical component of the revenue cycle. According to the American Medical Association, denial rate averages can be as high as high as 5% for Medicare. Rates vary by payer, but each payer requires different information and the overwhelming number of charges and claims make tracking denials impossible without comprehensive software.
While there are many solutions on the market today, many only show reports and “hot spots” for denials. These tools are not intelligent enough to predict which claims will be denied and how to change them before submission. More worrisome, many of these tools are developed and owned in part by payers themselves. There is little incentive to ramp up their effectiveness.
Denial management is ripe for machine learning to help providers get reimbursed what they are owed. A predictive system can highlight problems before they are submitted. By tracking the entire patient billing record from authorization through claims, an intelligent denial management system can flag problem areas and suggest interventions. Moreover, it can automatically augment pieces of the claim to reduce the burden for your team.
Like other machine learning tools, a predictive denial management software will also learn and adapt from the feedback of insurers. As rules and requirements change with new contracts, the system will automatically start to track and help your team react to those changes. Predictive denial management can drastically improve reimbursements, focus your team’s work on the most valuable areas, and reduce total workload.
AI tools to track Performance
Understanding performance and variation in the your hospital’s clinical care is key to driving change. Armed with better data, your administration can help negotiate better deals, guide clinical care decisions, and optimize care. Today’s machine learning tools can help your organization achieve change, improve care, and reduce costs.
5. Identifying medication variation
Your EHR holds hundreds of data points that can shine light on your care team’s clinical decisions including critical information about expenses. Medications are one of the fastest growing costs in hospitals; according to a report by the American Hospital Association drug costs increase an average of over 23% per year. Understandably, the pharmacy is often a major source of financial pressure for your hospital.
Many medications have equivalent effectiveness but vary widely in price. One example is oral Tylenol versus intravenous Tylenol. Creating and administering an IV for Tylenol can cost $15-35 or more per unit, whereas oral Tylenol costs pennies. Multiple studies show they are clinically equivalent (Journal of the Society for Academic Emergency Medicine, Journal of Hospital Pharmacy, Clinical Neurology).
So, how can you track variation in medication, determine levers of savings, and implement them in your organization?
Machine learning, again, can do the heavy lifting for your team. By analyzing patient profiles, prescribing physicians, medications, costs, and outcomes, these advanced systems can help your team make evidence-based decisions and drive down costs.
First, AI identifies what medications are driving costs and which physicians are prescribing them. This analysis enables your team to shift behavior to more cost effective drugs. Advocating and driving behavior change requires clear, insightful data. Ultimately, these tools will help your team save money and improve patient outcomes. An AI-powered medication variation software can deliver all this in one integrated solution.
6. Physician selection for capitation contracts
With the transition to value-based care, your provider group will have to negotiate contracts with payers. The shift is already happening, and value-based care will continue to grow dramatically over the next several years. CMS has stated they plan to move 50% of payments to value-based care by 2018.
In order to prepare for these opportunities, your group will have to understand how well you manage patient types and who is best-suited to engage with and keep those patients healthy. If there is not sufficient information, you will not be able to succeed in negotiating or managing costs under these contracts.
Fortunately, AI can arm your team with the data and insights you need to succeed in these changing environments. If you understand physician performance and variation by patient-type, you can select the best physicians to handle at-risk patients. The complexity of patient profiles makes this a difficult process without the tools to automatically understand, sort, and compare outcomes by physicians.
Moreover, these tools can help you guide behavior within your group to identify cost areas that can be managed. You will not only understand the best physicians to deal with certain patients, but also the best treatment plan that helped them heal quickly and cost effectively.
Final Thoughts and Call to Action
These applications and more are possible today by leveraging leading AI and data science firms in the healthcare space. By adopting these technologies, we can streamline and improve patient care while reducing costs for the entire healthcare system. HealthTensor, a company in Cedars-Sinai’s Accelerator powered by Techstars, is developing the core AI engine behind many clinical and operational applications for healthcare.