Healthcare faces ongoing struggles
The challenges faced by healthcare operations have been highlighted by the recent COVID-19 pandemic, and they reach more manageable peaks every year during the spread of seasonal illnesses like the flu and norovirus. Even during what is considered routine weeks and months in the healthcare sector, many facilities struggle to maintain adequate staff and consistency of care.
Implementation of new and improved systems in healthcare, meant to address these concerns, are typically extremely slow. They often occur too slowly, as in the case of COVID-19, to make any significant difference to the immediate situation. Although attempts are always being made to improve quality of care, retain the right staff, and make systems as consistent as possible, these efforts are usually not prioritized until they become emergencies, at which point it can be too late to impact current events.
One Japanese provider sought help
A provider of elder care in Japan maintains more than 400 care centers across the country. The Japanese government heavily funds their elder care efforts, providing citizens over the age of sixty-five with self-help resources, mutual aid, social support, and government-funded residential and therapeutic care, and this provider wanted to sink some of their funding into drastic improvements.
They sought to make their operations more efficient and less dependent on human interventions. Three key challenges were driving the investment:
- Difficulties hiring and retaining talent to care for the elderly. Due to the declining population of Japan, the high percentage of senior citizens, and the perceived dull nature of the work, young employees consistently left for other opportunities.
- Due to varying needs of the elderly within the same facility, managing resources to provide the right care without overspending was becoming difficult.
- The company had no daily decision-making system across facilities, so there was little consistency in care and staffing between locations.
The company decided to explore the potential benefits of AI. They wanted to work with a provider that could help them build a roadmap, implement the right solutions, and manage them on an ongoing basis. CrowdANALYTIX was chosen as the only provider that could meet their budget and provide the necessary end-to-end AI solutions platform.
Our approach
During a day-long discovery workshop, CrowdANALYTIX first mapped the business’ current operations process, identified key bottlenecks, and analyzed the state of structured and unstructured data stores. Next, a technical feasibility analysis of the data was performed and the company was delivered an AI roadmap with solutions separated into three horizons: short-term, near-term, and future. Among the potential AI solutions identified, CrowdANALYTIX began by working on improving the patient registration and room allocation processes.
Patient registration at speed
Upon registration patients are required to hand over all their belongings. This process takes 20-30 minutes, time that is frustrating for the care center executives and for the patients. CrowdANALYTIX built a solution to address this by building deep-learning models to recognize patient belongings, eliminating the need for manual input. Algorithms were built to structure the images and extract information like product type, color, pattern, and shape. This information was then automatically populated into a database. This saved more than half of the time previously required. This solution was supplied as a mobile application for use at the care center locations.
Room allocation process optimization
The room allocation process was ad hoc, based on spreadsheets. Because use of rooms was not optimized, many patients were refused due to lack of room availability despite many rooms standing empty. This was causing a significant revenue loss for the company. CrowdANALYTIX first developed a rules-based engine to automate the process of allocating rooms and optimize allocations for full room utilization. A full year was spent field-testing the algorithms and collecting additional data to convert the engine into a machine learning algorithm that could continually improve upon the process. The room allocation system was provided as a dashboard for care center use.
Impact
Patient registration times declined from 25-30 minutes to less than 15 minutes now, and are expected to continue their decline. The room allocation system has been implemented in 10 care centers, and there has been a 6% increase in utilization of rooms, a huge increase in revenue and margins.