In 2022, US health spending reached a whopping $4.5 trillion, as reported by the National Health Expenditure Accounts (NHEA). Despite such substantial investment, many healthcare providers still rely on outdated methods for record-keeping, with some using technologies developed over a decade ago or, in some cases, no technology at all. This gap in technological adoption for healthcare workers presents a double-edged sword: a challenge due to healthcare providers' unfamiliarity with modern tech, and an opportunity to leapfrog outdated systems with cutting-edge technology.
The healthcare sector involves detailed documentation and specialized terminology, areas that could benefit from enhanced standardization in decision-making processes.This environment is ripe for the integration of artificial intelligence systems, like “Retrieval Augmented Generation” and generative AI systems more broadly, provided it is implemented in a safe, ethical, and responsible manner. This article will explore two case studies that demonstrate how AI technology can significantly enhance operational efficiency in healthcare. By streamlining processes, these AI solutions not only aim to reduce costs but also enable providers to devote more resources to direct patient care rather than to indirect costs.
The first use case is specifically based on a real implementation for a skilled nursing facility, but could be applied to other providers that offer short-term or long-term care. This adaptable solution showcases the potential for AI systems to revolutionize patient admissions across various healthcare settings.
When hospitals are ready to discharge patients to a skilled nursing facility (SNF), finding the most appropriate SNF quickly is crucial. Hospitals typically send patient referrals to multiple SNFs within the area. The SNFs understand that a prompt response increases their chances of receiving the resident, which can positively impact their financial performance. However, the decision to accept a patient involves navigating through their extensive referral packet, which can be up to 100 pages and contains critical medical and personal information. SNFs must quickly yet accurately determine if they can meet the prospective resident's specific care needs without compromising the quality of care or incurring financial losses.
To address these challenges, we built the Nursing Home Admissions Co-Pilot, a “Retrieval Augmented Generation” tool (Generative AI with data) designed to streamline the admissions process. This tool rapidly processes the documentation associated with each patient referral, extracting key information necessary for making informed admissions decisions. By automating the analysis of detailed medical records, the AI tool assists admissions teams in quickly assessing whether a patient's needs align with the facility's capabilities.
The system is further enhanced by an admin portal that allows facility leaders to adjust the operational parameters of the AI tool. Leaders can set and update the criteria for medical condition alerts and care requirements, tailoring the tool to the facility's specific needs and specialties. This customization ensures that the AI tool remains adaptable and effective in meeting the dynamic requirements of patient care and facility management.
The deployment of this AI solution represents a significant improvement in how SNFs manage admissions, enabling quicker, more accurate decision-making and better alignment with the optimal care outcomes for patients.
This second use case centers around a smart data comparison tool developed for a provider’s billing department, aimed at enhancing billing accuracy in a fast-paced healthcare environment. The solution is crafted to address the common discrepancies in record-keeping and billing, ensuring financial stability for healthcare providers.
Healthcare providers operate at a high speed, making the maintenance of accurate records both critical and challenging. The stakes are high as over-billing is likely to trigger audits, and under-billing can undermine the financial health of the organization. In many instances, healthcare systems suffer from interoperability issues where different systems fail to communicate effectively. Some providers still maintain records on paper, creating further challenges when these need to be digitized and reconciled with electronic health records (EHR). These issues are compounded when modern tech solutions are implemented but fail to integrate seamlessly, introducing risks and inefficiencies into the billing process.
To mitigate these risks and streamline the billing process, we developed a smart data comparison tool that uses Generative AI models to intelligently match records, even in the absence of unique identifiers. For instance, the AI can recognize and reconcile variations in name order between datasets, such as "John Doe" in one and "Doe, John" in another, and can even assign temporary unique identifiers to facilitate accurate comparisons.
The tool allows billing teams and caregiving staff to upload their respective files—whether these are digital records from an EHR or digitized notes kept by providers. The AI then processes these files to match patients and compare data across different fields, highlighting discrepancies for further review. This automated comparison drastically reduces the hours previously spent manually matching data, which was not only tedious and prone to errors but also diverted caregiving staff from patient care.
By implementing this AI-driven tool, the billing department and caregiving teams can now focus their collaborative efforts on resolving only the identified discrepancies rather than reviewing every line item. This efficiency gain not only speeds up the billing process but also enhances accuracy and reduces the likelihood of financial discrepancies. It frees up significant time for caregiving staff, allowing them to dedicate more time to direct patient care, thereby improving both operational efficiency and patient outcomes.
This use case demonstrates the powerful role AI can play in solving complex, critical challenges in healthcare settings, enhancing not just financial operations but also the overall quality of care provided.
Both case studies benefited from three important technologies. Here, I'll share what they are, how to use them, and why they are so important. With these three components, savvy technologists can build extremely powerful and secure applications for healthcare in a short period of time.
AWS Textract is vital for extracting text, handwriting, and other data from various scanned documents, such as patient files and billing records. Utilizing advanced machine learning techniques, Textract exceeds the capabilities of traditional OCR by accurately capturing complex data formats efficiently, thus streamlining the subsequent data processing stages.
Credal is instrumental in maintaining the privacy and compliance of data within healthcare applications. Credal offers an Enterprise LLM Gateway product that anonymizes personally identifiable information (PII) to adhere to stringent regulatory standards. We enhanced its functionality by creating the "Credal Co-Pilot," a solution deployed via a REST API, which integrates directly into the data processing workflow to provide real-time anonymization.
Retool has proven invaluable in quickly developing custom user interfaces that are both functional and user-friendly. It allows for rapid customization and alignment with healthcare providers' existing IT ecosystems, including seamless integration with Microsoft Outlook for user authentication. This integration not only ensures secure and streamlined access management but also minimizes the maintenance and operational overhead, focusing resources on enhancing patient care rather than managing technology.
These components form the backbone of our technical architecture, enabling the rapid development and deployment of secure, compliant, and efficient healthcare applications that meet the dynamic needs of the industry.
With U.S. healthcare spending already substantial and projected to grow, the importance of integrating modern technology into healthcare operations cannot be overstated. This represents not only a challenge but also a tremendous opportunity to enhance the efficiency and quality of care provided.
To healthcare providers, this is a call to action to embrace AI and other advanced technologies. The shift from traditional methods to AI-enhanced processes is not just about keeping up with technological advancements but about fundamentally improving the way care is delivered. By reducing the time spent on administrative tasks, healthcare professionals can focus more on what they do best—caring for patients.
For developers and technologists, the healthcare industry presents a fertile ground for innovation. AI technologies are crucial unlocks for this field, providing tools that can handle complex data and improve operational efficiencies. I encourage technologists to consider healthcare as a critical area in desperate need for AI applications. By developing solutions that address the unique challenges of this industry, we can have a profound impact on the quality of care and operational effectiveness. Let's leverage our expertise to create systems that not only manage healthcare data more efficiently but also support providers in delivering better patient care. Together, we can transform the landscape of healthcare, making it smarter, more responsive, and ultimately more human-centric.
—---
This article is guest authored by Ilan Buckman, Founder and Principal of Raise the Bar AI, a consultancy dedicated to the healthcare sector. Raise the Bar AI specializes in identifying and crafting tailored solutions that leverage the power of AI to enhance healthcare operations and improve patient outcomes.
Credal gives you everything you need to supercharge your business using generative AI, securely.
