Artificial intelligence (AI) tools have reshaped the way hospices approach clinical documentation and compliance. However, the challenge, according to some hospices, is that the range of AI documentation tools currently available are limited.
Dr. Brian Haas, physician, founder and CEO of Wellspring Healthcare, said many AI documentation technologies are not specifically designed for hospices and are often more oriented toward other forms of home care. He founded the Missouri-based critical care provider in 2021.
Documentation errors are one of the most common red flags that result in regulatory scrutiny. This led Haas to form a company called Hospice Intelligence, which was incorporated in September 2025. A new artificial intelligence (AI) platform for hospice documentation is designed to guide clinical decision-making processes such as eligibility criteria, local coverage determination (LCD) of patient end-of-life status, and recertification.
“Hospice nurses provide excellent care, but the structure of hospice documentation makes it very difficult to consistently capture all LCD elements, especially when time is constrained,” Haas told Hospice News. “Rather than replacing clinical judgment, AI offers one of the greatest opportunities to date to support hospice clinicians by reducing cognitive load and enhancing documentation quality.”
Haas is a hospice physician and national medical director of hospice at Ascend Health. The health system provides hospice, home health and palliative care in Kansas, Massachusetts, Missouri, New Jersey, Pennsylvania, Virginia and Puerto Rico. Ascend Health also offers end-of-life doula services, advance care planning, cancer and memory care, home behavioral health counseling, programs for end-stage chronic obstructive pulmonary disease (COPD) and heart failure, and pet, massage and music therapy.
Avoid red flags
Hospices navigate a complex web of regulations, especially when it comes to clinical documentation, Haas noted.
Common documentation errors include insufficient documentation to justify a 6-month terminal prognosis, which is a major factor in recertification denials. The story should be specific, detailing the patient's condition and explaining why the patient is expected to reach the end of life within six months. This description must also be consistent with the LCD documentation issued by the Medicare Administrative Contractor (MAC) serving the hospice region.
One of the most common reasons hospices leverage AI tools is to streamline documentation. Documentation software is used for functions such as updating care plans, recording interdisciplinary team notes and patient visits, and tracking compliance and family communications.
Some AI tools have a lot of room for improvement when it comes to hospice documentation, Haas noted. These tools are not designed to interpret Medicare hospice eligibility, LCD criteria, comparative decline, certification period, prognostic indicators, etc., he said.
Additionally, AI tools are often not HIPAA compliant in their “native form,” Haas said.
“A truly hospice-specific platform must reflect both the clinical and regulatory realities of hospice care, something that general-purpose documentation tools can never achieve,” said Haas. “General-purpose AI models cannot reliably interpret hospice decline. General-purpose AI tools are powerful, but they are not hospice tools. And expecting an AI model to perform hospice-specific documentation breaks the technology.”
Regulatory oversight is increasing amidst program integrity issues in the hospice field. More than half of hospices nationwide are undergoing multiple types of audits at the same time, and more than three-quarters are experiencing an increase in targeted inspection and education (TPE) audits.
Haas said the audit experience revealed a “fundamental truth” about the difficult compliance environment hospices face.
Faris Flournoy, CEO of Flournoy Health Systems, said AI tools do not adequately document the person-centered experience essential to delivering hospice care.
Founded in 2007, Atlanta-based Flournoy Health Systems is the parent company of PrimeCare Home Care Services, which provides hospice, home care and chronic care management to 59 counties in Georgia and Indiana. The organization's other services include pediatric skilled and unskilled nursing care in the home and applied behavioral analysis (ABA) therapy for children.
Compliance concerns are growing as advances in AI lag behind recognizing the nuances of hospice regulation evolution, Flournova noted. A critical challenge concerns the ability to document various points along a patient's end-of-life journey, a key aspect of the Hospice Outcomes and Patient Evaluation (HOPE) tool, a new quality reporting system introduced this year.
“We recognize that AI is not broad-based and requires building language models to capture data specific to hospice patients,” Flournoy told Hospice News. “We are missing a large part of the unique, human-centered experience at the end of life. This model must include understanding prognosis and diagnosis, and it must not rely on open source. The toolset fails in several areas and does not provide documentation support to nurses.”
Avoid compliance risks
According to Chad Hanner, senior vice president of products at nVoq Inc., AI tools should support individualized care plans and help clinicians identify gaps in documentation.
Colorado-based nVoq's documentation software is designed for home health and hospice providers. Among the company's tools is Note Assist, which helps identify missing eligibility evidence, defensible language that is vague or noncompliant, discrepancies between visit documentation, and care plan discrepancies that do not reflect current patient symptoms.
Hainer said AI tools need to go beyond transcription and get better at identifying clinically meaningful details and supporting a clear and defensible physician narrative.
Hospice AI documentation tools need to be grounded in the realities of care delivery, he said. Heiner said the technology should be designed around aspects of end-of-life care, such as patient and family privacy in the home environment, mobile clinical workflows, and reduced after-hours documentation.
Incorporating strong clinical governance into AI documentation is essential to ensure accuracy, reduce clinical bias, and maintain compliance, he said. These tools should be designed to increase efficiency, reduce documentation burden, uncover billing errors and improve pre-billing quality assurance clinical workflows, Heiner said.
“When thoughtfully designed, these tools support clinicians as active participants, focusing on the patient rather than the medical record,” Heiner told Hospice News via email. “Current models require a deeper understanding of hospices' clinical workflows, regulatory expectations, and interdisciplinary care. AI can serve as an early safety net in flagging issues. By discovering these gaps early, especially when it comes to decline indicators and recertification narratives, AI can help reduce compliance risk.”
Haas said one area of risk in hospice documentation is physician narratives. Some AI tools currently available generate clinically sophisticated narratives, but are not “audit-ready,” he said.
Hospice-specific AI documentation tools can help improve clinical accuracy and strengthen consistency across multiple geographic locations, reducing the variability that often triggers audits, Haas said. AI can greatly enhance audit readiness, but only if it's designed around Medicare's actual evaluation criteria, he said.
The most important key to hospice AI documentation is the human element, he said. Healthcare providers need tools that can accurately capture not only worsening conditions, but also the evolution of end-of-life care delivery, patient needs, and family preferences.
“Innovation in this area requires close collaboration between clinicians, compliance professionals, and technology developers, otherwise AI will not meet the true needs of hospice teams,” Haas told Hospice News. “AI should guide, but not dictate. It should preserve clinician judgment.”
According to Flournoy, ethical and practical considerations are important factors in the outlook for hospice AI documentation. He explained that human interaction is key to building AI capabilities with different reference points for data bias and understanding end-of-life care.
Hospices need strong quality and compliance oversight measures to ensure that AI tools augment, rather than replace, clinical expertise, Flournoy said.
“What are the checks and balances needed to incorporate AI-specific technology that loops clinicians in?” Flournoy said. “We need to be able to enhance or speed up certain processes and have better quality compliance support. I think both patients and healthcare providers will certainly feel more comfortable knowing that they not only have this speed and agility in the potential opportunities of AI, but they also have the talent to make them more compliant and more human-centered.”
