Going to ask again about HealWell - they are on an acquisition tear and seem to be very AI-focused. Has…
Readers Write: Applying AI to Improve Patient Care
Applying AI to Improve Patient Care
By Tomas Gogar
Tomas Gogar, MS is co-founder and CEO of Rossum of London, England.
Despite the technological advancements in healthcare over the past decade, the administration and quality of patient care has not kept pace. The industry is faced with the realization that if technological changes aren’t implemented at a foundational level, providers, payers, and patients won’t be able to realize the full value of the technology available to them.
The majority of medical institutions rely on electronic health records (EHR) to input, read, and upload critical documents related to patient care into online portals. The EHR concept, introduced in the 1960s, while valuable to the healthcare community, has yet to eliminate the need for manual paperwork. Paperwork is a huge drain and cost, taking time, energy, and precise attention to detail to ensure that all documents are properly scanned into the correct patient files.
Missing information can lead to delays in care, misdiagnosis, miscommunication around treatment plans, and the duplication of costly tests and procedures. Relying strictly on manual processes to manage such large amounts of information can be administratively crippling to a healthcare organization. The World Health Organization estimates that up to 50% of all medical documentation mistakes result from administrative errors.
By integrating intelligent document processing (IDP) into the systems, hospitals and healthcare institutions save time, reduce operational costs, and improve workflows. Introducing an IDP system into the EHR workflow means medical professionals across departments can easily scan and upload documentation into a secure SOC 2 and HIPAA compliant operating system. IDP efficiently captures, categorizes, extracts, and classifies data from documents, streamlining the workflow process and reducing the paperwork necessary for a patient file.
IDP also helps sustain HIPAA compliance, which can be challenging when dealing with thousands of physical documents stored in different formats and locations across a health system. Accounting for small margins for human error causes long input times and exhaustive efforts to safeguard physical documents containing patient information. With the implementation of IDP, this process eliminates any chance of human error in handling sensitive information and allows for patient data to be processed quickly, safely, and securely.
From a patient perspective, automating and streamlining document processing enables providers to get complete, accurate data straight into a patient’s hands via online portals. From the healthcare organization side, IDP can reduce document burnout that healthcare professionals are prone to experiencing.
For hospitals struggling with overhead operational costs, implementing IDP is a lucrative resource. By using IDP to process documents like prescription referrals, lab records, billing, and claims forms, manual data entry is drastically reduced, thereby reducing the need for resources associated with data entry into EHR and patient portals and enabling the healthcare organization to re-allocate them to more strategic tasks. In addition to labor costs, implementing IDP reduces costs associated with paper storage, security measures in place to store these documents, and any costs associated with administrative errors.
During a time when all our hospitals are critically understaffed and underfunded, ensuring that every worker is given the necessary tools and resources to adequately and efficiently perform their jobs is more crucial than ever.
“The EHR concept, introduced in the 1960s”
Dang, haven’t seen an updated version of the “MUMPS is old” trope or the “CDA has been around too long” message in a while.
Would you mind sharing where you think there is a need in healthcare for something like this? In my experience especially in billing paper is becoming obsolete. There are a few payors that aren’t able to send electronic remittances but those tend to be very small groups. Perhaps in smaller independent doctor offices that aren’t able to setup standard interfaces to accept things like electronic prescriptions this might be helpful?
Sorry, but we have to solve an awful lot of problems with data before we get to IDP, or any other form of “AI” in healthcare.
An article on those problems, and how you propose to solve them would be much better received. The data are bad, the data are not specific enough, and the data are not relevant to the patient’s care or treatment. With bad, incorrect, templated data the information created from it to train the AI is bad.
We have clinician notes that are templates without adjustments made to the template to ‘personalize’ it to the patient. We have notes that are repetitive, copied encounter to encounter or round to round without change.
We have EHRs that are way too clicky and frankly make searching for specificity difficult. You can create “favorite” lists but then how is that going to personalize the medicine to your patient. We have EHRs that use incorrect enumerations and code sets for their domains. The code sets are the thing that allow us to ‘organize’ and create the taxonomy needed to walk the DUR.
Lastly, how much of our data are from other sources, and I repeat, yet again, the sources do not merge data between them. Their interoperability is not only not yet proven but is demonstrably not functioning.