How can AI help predict, automate hospital discharge?
When planning discharge, care teams must use their best judgment and available data to inform decision-making. Can artificial intelligence tools streamline this process?
Discharge planning plays a pivotal role in transitioning patients from one level of care to the next. This process requires patients, caregivers and providers to create an individualized plan to ensure a smooth journey along the care continuum after leaving the hospital.
Planning discharge requires stakeholders to effectively take into account the patient's care needs, concerns and preferences. Doing so is key to helping patients stay healthy following hospitalization and reducing potential readmissions, as discharge delays can put a strain on both patients and hospitals.
But prioritizing continuity of care to reduce adverse outcomes necessitates the use of both clinical judgment and patient data. Doing so can present unique challenges for already overburdened care teams.
In a recent interview, Jean Halpin, COO at Grant Medical Center, discussed how the organization has deployed AI tools to help predict discharge dates and automate aspects of the discharge planning process.
The challenges of effective discharge planning
Halpin noted that a streamlined discharge process is key to reducing wait times and improving patient engagement, but multiple factors determine how quickly patients are discharged.
She emphasized that this challenge is particularly evident in the context of extensive emergency room (ER) wait times.
"When you pull back the curtain, most of the wait time we experience as patients boils down to a lengthy discharge process that isn't effectively moving patients," Halpin explained. "It's a domino effect -- the person waiting in the ER for an open bed is waiting on someone in the ER to be admitted for longer-term care, who's being delayed admission due to another admitted patient upstairs not being discharged when they should have been."
To optimize patient flows, the health system deployed Qventus Inpatient Solution, which integrates into EHRs and pulls patient data -- including clinical notes, patient history and labs -- to make recommendations to the care team regarding the best timing for patient discharge.
Halpin indicated that the inefficiencies in the discharge process surfaced by the tool have helped the health system improve patient flows, which has sped up the care coordination process and enabled more accurate discharge day predictions.
These insights have led to reduced wait times and enhanced care flows across Grant Medical Center's ERs.
Integrating discharge planning AI into clinical workflows
For health systems looking to take advantage of AI and other tools, integration into clinical workflows can present a challenge. Part of the integration process involves ensuring that a tool meets clinicians' needs without creating additional burdens or workflow hurdles.
By integrating with EHRs, the Inpatient Solution tool presents minimal workflow disruptions while providing valuable insights to care teams.
Halpin indicated that once the tool pulls the necessary EHR data for a patient, it processes that data to generate an estimated date of discharge and a summary that clinicians can reference as they provide care.
"As patient health changes, the [discharge] date can fluctuate as well, but the AI is able to use its knowledge to predict the most accurate day based on similar patient cases. The care teams can then look at the date, and note whether or not they agree, without having to dig through patient history and research other cases to develop their own recommendation," she said.
Halpin also underscored the tool's value for reducing administrative burden.
Health system leaders should embrace the advancements available in the industry that can help alleviate burdens for workers, allowing them to get back to working at the top of their license … Once they fully understand, they can feel empowered to prioritize clinical interactions and care with patients while AI can handle the more time-consuming admin tasks.
Jean HalpinCOO, Grant Medical Center
"Tasks like discharge coordination to rehab facilities, the ordering of tests, prescription of medication and more, consume healthcare teams' time," she stated, noting that Inpatient Solution has helped streamline or automate these tasks. "After all, the more time our care teams are digging behind the screen for answers, the less time they can spend with patients."
To date, the teams at Grant Medical Center most positively impacted by the tool's adoption are the physical therapy, imaging, and lab teams. Halpin stated that these teams can reference Inpatient Solution's recommendations within patient charts to determine which patients should be prioritized for testing or which ones are ready to go home and come back as outpatients.
However, deploying any new tool in a hospital comes with growing pains and bumps in the road.
"It takes time to fully integrate [AI] and get teams up to speed on best practices to avoid any hiccups in workflow," Halpin explained. "Qventus was able to onboard all of our teams over an extended period of time, allowing the opportunity for training, questions, and clarifications, which made the transition period much more seamless."
Embracing AI to alleviate healthcare worker burden
Alongside addressing integration concerns, health systems adopting AI to enhance hospital discharge must accurately assess the tool's efficacy.
At Grant Medical Center, leadership is looking to employee satisfaction, patient outcomes and administrative improvements -- such as time and money saved -- to gauge success.
"By tackling the gaps in our patient [flow] and workflow, we were able to expedite the speed of care, getting patients admitted earlier to be seen and out the door once they were ready to go home, accounting for a reduction in excess stays for patients by nearly 1,400 days. Patients are happy to go home in a timely manner, and our care teams are happy to be back with the patients working at the top of their license," Halpin said.
But she highlighted that successfully piloting AI for a use case like discharge planning requires stakeholders to balance the hype and concerns around these tools.
"Health system leaders should embrace the advancements available in the industry that can help alleviate burdens for workers, allowing them to get back to working at the top of their license," she said, noting that while staff might be initially hesitant about using AI, working hand-in-hand with care teams to onboard these tools can make the transition easier and help them better understand relevant use cases.
"Once they fully understand, they can feel empowered to prioritize clinical interactions and care with patients while AI can handle the more time-consuming admin tasks," Halpin concluded.
Shania Kennedy has been covering news related to health IT and analytics since 2022.
