Home | About | Events | Contact | Advisory Board | Writer’s Guidelines | Reprints | Organizations | Advertising
Summer 2025
Summer 2025 Issue Data-Driven Elder Care A look at how predictive analytics is shaping personalized care plans. As technology continues to transform health care, data-driven approaches are revolutionizing elder care. Predictive analytics are now at the forefront, enabling providers to craft personalized care plans that anticipate older adults’ needs and improve outcomes. “Predictive analytics is a gamechanger for value-based care environments, specifically those tied to Medicare and senior care,” notes Moti Gamburd, CEO of CARE Homecare, a Southern California home care agency specializing in Alzheimer’s and dementia care. “It allows care teams to support seniors with multiple chronic conditions and complex needs to make the shift from a reactive approach to more proactive, personalized care strategies.” Predictive analytics allows providers, care teams, insurers, and the like to understand patient cohorts, segmenting populations based on risk trajectories, care gaps, and behavioral patterns. So much so that employing predictive analytics technology has been pivotal in advancing care delivery for older adults in supporting a more personalized plan of care and ultimately enhancing their quality of life. “We can now take the large amount of information contained in the patient’s medical record to leverage more advanced treatments, recommendations, and earlier referrals,” explains Kurt Merkelz, MD, senior vice president and chief medical officer of Compassus, a provider of hospice, palliative, and home health care services. “Information such as past history, diagnosis, complications, laboratory data, vital changes, weight changes, medication use, and changes all can be pooled and analyzed to predict the probable patient journey.” Additionally, wearable sensor data and even social determinants of health information can be pooled, providing even more accurate prediction models that can then be used for targeted interventions. Kate Martin, MD, MEd, MPH, FAAFP, a professor of family and community medicine at University of Nevada, Las Vegas’s medical school, sees predictive analytics moving beyond high-level risk scores to actionable, person-level care plans. “For example, our EHR systems now use machine learning to monitor vital signs, lab results, and nursing notes for patterns that can be concerning for a worsening prognosis, ie, sepsis, and can help clinicians in decision-making with regard to who is at higher risk for readmission to the hospital,” she says. Jeremy Clerc, founder and CEO of Assisted Living Magazine, an online senior living resource and marketplace partnering with more than 4,500 senior living operators, notes predictive analytics is fundamentally shifting elder care to a proactive rather than reactive, highly personalized model. “It enables that by drawing on huge reservoirs of historical and current data to predict individual needs and health risks in advance before they become crises,” he says. Among the most important clinical use cases he’s seeing are early health deterioration detection, creating personalized treatment plans, and optimizing resource allocation, critical for operational efficiency. “Facilities can use predictive models to forecast staffing needs, manage supplies, and streamline patient flow, so that the right resources are present at the right time,” Clerc says. “These applications are not only improving outcomes, they are totally revolutionizing the paradigm of elder care.” Rise of AI “Using large language model (LLM) reasoning, AI can help family members and personal care aides break down often complex problems into smaller, more manageable steps,” Martin says. “For example, a LLM system can ask caregivers a series of questions to develop goal-oriented plans in areas such as sleep, mobility, and socialization.” AI also plays a very important and supportive role in care planning for patients with dementia or cognitive impairment, where personalization is crucial but nuanced. “Cognitively impaired patients require very unique and individualized care plans,” Merkelz says. “Leveraging information to ensure that the most appropriate treatment is provided at the earliest and most effective time is crucial. The information can help support longer home stays, potentially delaying more costly caregiving options. Many medications, perhaps appropriate at an early stage, can be timely planned for deprescribing when the actual risks may outweigh the benefits conferred.” Analysis in Action “AI technologies analyze data from wearable sensors (accelerometers, gyroscopes), gait-tracking systems, or even vision-based systems to identify abnormal patterns of movement, instability of balance, or muscle weakness that usually preclude falls,” Clerc says. “Predictive warnings enable caregivers to implement preventative interventions like physiotherapy or home modifications, significantly reducing the incidence of falls and injuries.” Eric Jutkowitz, PhD, an associate professor at the Brown School of Public Health and founder and CEO of Plans-4Care, which provides personalized dementia care solutions, notes that predictive analytics is utilized for the Plans-4Care’s Digital Caregiver Playbook. “The Caregiver Playbook includes the Everyday Function Scale, which identifies the abilities of the person living with dementia and provides caregiver-friendly descriptions of these abilities,” he explains. “The Caregiver Playbook also generates personalized care plans with strategies to address dementia-related care challenges, tailored to the individual’s specific capabilities.” Chronic disease exacerbation and hospital avoidance are two of the most impactful areas where pooled data analysis can best be seen, according to Merkelz. “By looking at specific patient data, subtle changes can be seen to best leverage more timely referrals and earlier treatment responses,” he says. “Patient-specific data can help guide the referral process, ensuring the right care is offered at the right time. For the patient, this predictive modeling can support earlier access to resources and treatments, which can best support their individual health care journey.” After all, early treatment algorithms for chronic heart and lung disease allow for earlier detection of changes so increased treatment responses can be offered, reducing the likelihood of acute exacerbations, decompensation, and hospitalization. In home health, predictive modeling can be used to provide a more integrated solution of treatment options, recommending earlier hospice transitions, and avoiding unnecessary hospitalizations that often accompany a patient’s final days and weeks of life. “In hospice, we leverage this data to help ensure patients whose combination of changes, which in frail older adults may be more subtle—weight, vitals, oral intake, behavior, medication usage, etc—to help identify patients who may benefit from increased clinical oversight to ensure patients have increased attention during their final days,” Merkelz says. Martin notes that machine learning algorithms better help detect polypharmacy (multiple medications prescribed to the same person) in older adults and alert health care providers to patterns associated with falls and cognitive issues. “No one wants an older adult patient to experience a bad outcome due to the additive or conflicting effects of multiple medications, but many don’t have time to locate and review all of the medications that are being prescribed, typically from several specialists,” she says. “I have seen older patients who were taking multiple versions of the same blood pressure medication that was prescribed over time by different physicians. The sooner a problem like this is detected, such as through the benefit of AI, the less chance of adverse effects occurring over time.” Brandon Blakeley, a senior care professional at Mirador Living, has seen predictive tools being used to analyze a wide range of data, from EHRs, medication usage patterns, and prior hospitalizations to behavioral trends and even data from wearable devices like Fitbits or smart fall detectors. “These technologies can flag warning signs well before a visible issue arises,” he says. “For instance, if a resident’s movement decreases over several days, or if sleep patterns suddenly shift, predictive models might alert caregivers to an increased risk of a urinary tract infection or early-stage pneumonia—both of which are common and dangerous in older adults. This early warning enables intervention before hospitalization becomes necessary.” One specific example comes from a memory care community he advised in Florida, where predictive analytics helped identify residents who were at risk of elopement—a dangerous scenario where an older adult with dementia wanders off. “The system flagged behavior patterns such as increased restlessness in the evening or multiple attempts to leave their rooms, allowing staff to put additional safety measures in place before an incident occurred,” Blakeley says. Reducing Burden “Automating documentation has been a major enhancement,” Merkelz says. “The time commitment required for this regulatory requirement is immense. Utilizing AI-driven technology, tools are now available to summarize and scribe the clinician-patient interaction and even auto-generate notes that the clinician needs only review and authenticate, saving a significant amount of time and labor as well as supporting more thorough, accurate, and patient-detailed documentation.” What’s more, this detailed documentation can further support better care plans and facilitate better interdisciplinary communication and thus patient care, reducing time spent laboring through countless pages of documentation to find important information about the patient. “Detailed information, patient and facility dashboards allow for more patient-specific care protocols and care plans to meet the unique needs of each patient,” Merkelz says. “And predictive modeling in the long term care facility can allow for more timely interventions to reduce infections, pressure ulcers, and weight loss, as well as falls.” Jutkowitz warns that care planning technology should not replace human support; rather, it should enhance the process and extend the capabilities of staff. Data Matters “Predictive analytics is hungry for deep, diverse data sets,” Clerc shares. “AI needs to transform difficult-to-interpret behavior and health data into simple-to-interpret actionable insights for caregivers so that they can gain a better understanding and respond to residents’ needs in a manner previously dependent on sheer intuition.” Thankfully, there is a huge ecosystem of data available, including wearable devices, remote patient monitoring, EMRs, claims data, long term care assessments, patient-reported outcome measures, social determinants of health, demographic data, and more, all of which can be pooled to leverage more appropriate and timely care. Discounting Bias Merkelz notes this is a vital area still being addressed today. After all, when AI models learn from data that does not accurately represent diverse populations of older adults—eg, various ethnic groups, socioeconomic levels, or those with complex comorbidities—they can reinforce bias and downstream bad predictions, misdiagnoses, and unequal treatment. “It’s important that the data used to generate these models is well representative of populations,” he says. “The models will require continuous quality assessment and improvements, ensuring they reflect performance across all individuals.” Denise M. Brown, founder and CEO of Caregiving Years Training Academy, an online research and training program for caregivers, believes it’s the personalized coaching that prevents bias, and it’s the reason trained peer support can be such a unique and effective intervention. “The trained peer support professional has a unique perspective which does not include bias,” she says. “When the trained peer support professional understands the unique needs of the client, then the professional can ensure predictive models provide appropriate suggestions and instructions. The client doesn’t receive a general plan of care. Instead, the professional develops a strategy that takes into account a client’s specific challenges and needs.” Using predictive models saves the professional time in developing the care plan because the model can generate evidence-based strategies from which the professional can choose. The professional then develops personalized strategies clients can apply to their particular situation, which also saves the coach and the client time. Challenges to Overcome “There will be much to learn as continued adoption takes place,” Merkelz says. “There are still cultural barriers, privacy issues, and significant interoperability issues that need to be addressed. Mistrust of technology, ensuring safe handling of patient-specific data, consents, resistance to change—there is still much to overcome.” For health care systems looking to dive deeper into predictive analytics, Merkelz offers some savvy advice: “Start small and look for opportunities to improve workforce improvements where current processes hamper good care delivery, such as improving documentation and reducing caregiver burnout,” he says. “Focus on the patient. So much time is spent documenting care that the real needs of the patient can become lost. Look for ways to improve information exchange and leverage the information you already have to drive more accurate care plans.” As more digital tools are used to provide services to older adults, Jutkowitz notes that a greater amount of data will be collected that could help inform predictive modeling. “However, these tools often do not collect all the data needed to make meaningful predictions,” he says. “The challenge moving forward will be to link data between sources to generate accurate and insightful predictions.” Looking Ahead “Wearable devices, remote monitoring, and utilization of predictive models will improve at a rapid rate, likely influencing every facet of our lives,” Merkelz says. “It seems at this point for the better, tailoring specific recommendations, plans, and treatments based on an individual’s past and present condition. It can lead to safer living environments, more convenience, more efficient and accurate care delivery, and tailored medication treatments with fewer side effects and adverse events.” Clerc sees an integration of smart home, Internet of Things, and AI technologies accelerating in the upcoming years. “These integrated homes will constantly monitor activity levels, sleep patterns, and environmental conditions, modifying the living environment in real time and detecting hazardous situations,” he says. “Ubiquitous monitoring will drive more accurate predictive models, further enhancing safety and independence for seniors aging in place.” Martin is excited about the future of predictive analytics within senior care, particularly at the intersection of aging and disability in home-based and community settings. “Home care agencies equipped with machine learning system software can help predict the need for virtual vs in-person doctor visits, allowing for a triage process that can efficiently leverage health care resources based on clinical algorithms,” she says. “We will enjoy more products that can support an older adult with mild cognitive impairment in taking their medications as scheduled through reminders that are personalized to their particular clinical profile.” — Keith Loria is a freelance writer based in Oakton, Virginia. |
