A systematic review on the efficacy of artificial intelligence in geriatric healthcare: a critical analysis of current literature

Abstract

Objective

To carry out systematic analysis of existing literature on role of Artificial Intelligence in geriatric patient healthcare.

Methods

A detailed online search was carried out using search phrases in reliable sources of information like Pubmed database, Embase database, Ovid database, Global Health database, PsycINFO, and Web of Science. Study specific information was gathered, including the organisation, year of publication, nation, setting, design of the research, information about population, size of study sample, group dynamics, eligibility and exclusion requirements, information about intervention, duration of exposure to the intervention , comparators, details of outcome measures, scheduling of evaluations, and consequences. After information gathering, the reviewers gathered to discuss any differences.

Results

Thirty-one studies were finally selected for systemic review. Although there was some disagreement on the acceptance of AI-enhanced treatments in LTC settings, this review indicated that there was little consensus about the efficacy of those initiatives for older individuals. Social robots have been shown to increase social interaction and mood, but the data was more conflicting and less definitive for the other innovations and consequences. The majority of research evaluated a variety of results, which made it impossible to synthesise them in a meaningful way and prevented a meta-analysis. In addition, many studies have moderate to severe bias risks due to underpowered design

Conclusion

It is challenging to determine whether AI supplemented technologies for geriatric patients are significantly beneficial. Although some encouraging findings were made, more study is required.

Background

The ageing population around the world is dramatically rising, posing a threat to the viability of conventional approaches of healthcare that have centered on in-person surveillance. An extended lifespan frequently entails dealing with disabilities and chronic illnesses that can make it difficult for people to carry out daily tasks or operate independently [1,2,3]. Many nations are struggling with a distinct lack of direct care professionals like home healthcare workers, which adds to the pressure from an ageing population that needs increased levels of individualised attention, support, and care. The difficulty of replacing the ageing health workforce persists. Women make up the majority of informal carers worldwide, and they frequently juggle caring for their elderly relatives while also fulfilling other domestic, familial, and work obligations [4, 5].

The possible number of family carers per elderly person is also anticipated to decline significantly as a consequence of shifting family dynamics, shrinking family sizes, women’s increasing engagement in the employment, and migration trends [6, 7]. Many senior citizens place a high importance on living independently or “ageing in place,” which implies they want to remain in their current residence with the necessary support instead of enter institutional care, which is similarly in limited supply and may be out of reach for many senior citizens [8, 9].

The recent COVID-19 global epidemic, which has profoundly influenced older individuals, particularly those already in long-term care centres (LTCs), strengthens calls for alternative approache to support people in staying as independently as possible in their homes and/or receiving continuous monitoring of health that needs the smallest amount of face-to-face contact. Systems that are using video cameras for recording people’s behaviors at home as a part of remote surveillance systems, could help older adults maintain their independence. These technologies however still depend on human workers or families and carers to be observing video streams in real-time and acting in accordance with their judgements. As a result, they require a lot of labor and are vulnerable to human mistakes and diversions [10, 11].

Will emerging computerized and ongoing innovations, like Artificial Intelligence (AI) healthcare monitoring, improve older folks’ capacity to function safely in their preferred circumstances as we face mounting human resource hurdles?

There has not yet been a thorough synthesis and evaluation of the research on AI-enhanced approaches in LTC services for geriatric patients. The generalizability of outcomes and clinical consequences may be impacted by existing assessments that have concentrated on other technological solutions like environmental sensors and social robots for supporting older people, but not focused over LTC, [7, 8]. The comparisons with certain other AI-enhanced approaches that might be more approachable and less expensive is limited because prior assessments have concentrated on individual robots (like PARO) in senior care facilities [9, 12]. We conducted a thorough literature assessment on the tolerability and efficacy of AI-enhanced treatments for senior citizens receiving LTC in search of a solution to this lack of evidence.. The purpose of this systematic review is to respond to the following research queries: (1) What LTC services-related AI-enhanced treatments have been tested? (2) Which AI-enhanced activities for senior citizens undergoing LTC have been proven to be successful? (3) Which AI-enhanced initiatives have been demonstrated to be well-received by senior citizens undergoing LTC?

Methods

Eligibility criteria

The eligibility criteria are presented as –

Criteria	Inclusion	Exclusion
Type of Publication	Journal papers with peer review and recognized comprehensive conference proceedings	Non-peer-reviewed papers, theses, and dissertations
Research Type	- Pilot investigations - Viability and acceptance research - Controlled (non-randomized) clinical research - Pre-evaluation and post-evaluation clinical research	Opinion pieces- Guidelines - Laboratory tests- Qualitative research - Literature reviews
Participant Age	Senior citizens with an average age of 65 years or higher	Study groups with an average age under 65
Technology	AI-based systems for monitoring and/or delivering healthcare interventions	Research that did not use an AI method
Outcome Focus	Technology aimed at promoting results related to physical ability, intellectual acuity, physical well-being, or mental well-being	Research that did not assess acceptance or the impact of the AI technology on a clinical outcome and merely evaluated sensibility, precision, or efficiency
AI Approaches	Including but not limited to pattern recognition, machine learning, natural language processing, automation systems, and robotics
Control Group	Comparators could include therapy as usual, waitlist management, a substitute AI, or an active control without AI
Evaluated Outcomes	Studies evaluating acceptability as an outcome, or other health results related to physical well-being, mental well-being, or emotional well-being
Healthcare Setting	Long-term care homes for geriatric patients, community creche centers for geriatric patients, or home-associated long-term care centers	Research where the AI technology wasn't assessed in a long-term healthcare setting
Physical Form of Technology	The physical form of the technology did not disqualify studies

Search strategy

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) standards were followed in the preparation of the review (Fig. 1). With the exception of an upgraded search method that increased the specificity of findings, there were no protocol violations. Using search phrases including artificial intelligence (AI), robotics, environmental sensors, wearable sensors, social robots, long-term care services, geriatric patients, acceptance, etc., a thorough search was carried out on Pubmed database, Embase database, Ovid database, Global Health database, PsycINFO, and Web of Science. Dates of publishing were not restricted. On March 31, 2023, systematic searches were carried out.

Fig. 1

PRISMA flow diagram

We developed two sets of keywords to encompass the eligibility criteria. The first set of keywords used “artificial intelligence” OR “machine learning” OR “deep learning” injunction with an AND operator and the terms “elderly patients” OR “older adults.” The second set of keywords consisted of MeSH terms using the PubMed MeSH database. The used terms were “disease attributes,” “aged,” and “artificial intelligence.” The MeSH term “disease attributes” was used and then sorted to ensure all articles, including chronic illnesses along with the other keywords, were included. The term included topics such as “acute disease,” “asymptomatic diseases,” “catastrophic illness,” “chronic disease,” “convalescence,” “critical illness,” “disease progression,” “disease resistance,” “disease sus ceptibility,” “diseases in twins,” “emergencies,” “facies,” “iatrogenic disease,” “late-onset disorders,” “neglected diseases,” “rare diseases,” and “recurrence.” “MeSH terms are organized in a tree-like hierarchy, with more specific (narrower) terms arranged beneath broader terms. By de fault, PubMed includes in the search all narrower terms; this is called “exploding” the MeSH term. Moreover, the inclusion of MeSHterms optimizes the search strategy. All of the articles that fit our inclusion criterion were analyzed to make sure the disease they were targeting was chronic before continuing forward”.

The bibliography of pertinent systematic reviews and the publications of the included authors were also manually searched.

Data collection

The data was retrieved in sequence into separate worksheet formats by two impartial reviewers (AB, CDE). Study specific information was gathered, including the organisation, year of publication, nation, setting, design of the research, information about population, size of study sample, group dynamics, eligibility and exclusion requirements, information about intervention, duration of exposure to the intervention, comparators, details of outcome measures , scheduling of evaluations, and consequences. After information gathering, the reviewers gathered to discuss any differences.

Risk of bias assessment

To determine the internal reliability of the included publications, the possibility of bias was assessed at the level of research. To settle disagreements, reviewers gathered with a third research group member (RR). Due to the significant degree of variation in the results and metrics used across research, a descriptive summation was carried out under the direction of Popay and colleagues. 12 publication characteristics were tallied and descriptively reported. Depending on whether a control category was included in the research, the findings were shown and graded. For each kind of investigation design, the patient outcomes which have been evaluated in more than 3 papers were enumerated. A narrative format was used to report the findings as a whole.

Data analysis

κ statistics

Using the kappa (κ) statistic, we determined the level of consensus among the two evaluators for the determination of admissibility. The disparity between the two research teams in the change in rating from pre to post therapy was the main summary statistic we took into account in determining the success of the therapy. Since, the included publications either provided no statistical data or provided different summary statistics, we decided against doing the proposed meta-analysis.

Results

Results of literature search

Two hundred fifty-two papers were obtained through literature search by using search terms. 160 similar and duplicate papers were excluded. 92 distinct articles were selected initially. 43 articles excluded after reviewing abstracts and titles. 49 papers selected for which full text was managed. 06 extra papers found manually from references. 55 articles with full text were eligible for study. 24 inadequate articles excluded in final. 31 studies were finally selected for systemic review (Fig. 1 and Table 1)

Table 1 Salient features of all studies included in the systematic review

Overview of included Studies

The studies included in this study were conducted between 2004 and 2021. The studies were conducted in countries like USA, New Zealand, Australia, Norway, Canada, Denmarch, Spain, Switzerland, Taiwan, Greece etc. The interventions used in studies were environmental sensor devices [24, 25, 39, 40], wearable sensors [15, 29, 37, 38]. These sensors were used in some studies for recording and monitoring health status. The robotic systems were found to record the vital signs and provide cognitive games, video callings and entertainment videos [13, 14, 16,17,18,19,20,21,22,23, 26,27,28, 30,31,32,33,34,35,36, 42, 43]. The controls used in the studies were living dog [13], standard care [14,15,16,17,18], plush toy cat [19]. The number of study participants varied from 4 to 415. The study settings included LTCs, nursing homes, dementia day care centres, living communities, residential care hospitals. Mean age of study participants in different studies varied from 67 years to 98 years. The average duration of exposure to intervention devices in studies varied from 30 minutes to 24 by 7 in different studies. The time intervals which were taken as reference point for follow up included 4 week follow up and it extended upto 2 years in some studies. The outcomes that were analysed in different studies were depression [14, 16, 20, 23,24,25,26, 30, 32, 34], quality of life [14, 17, 20, 26, 28, 30, 32, 34], agitation [16, 18,19,20, 23].

Methodological quality of included studies

The studies included were controlled trials [13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29] and non controlled trials [30,31,32,33,34,35,36,37,38,39,40,41,42,43]. Among controlled trials there were randomized controlled trials [13, 27, 28], non randomized trial [14, 25], pilot RCT [15, 18, 19, 23], retrospective study [29]. Tables 1 and 2.

Table 2 Quantitative results of studies showing the important results and conclusions with different AI tools

Reporting biases

Some studies showed low risk of bias [16, 18, 21, 24, 26,27,28,29,30, 32, 36] while some studies had high risk of bias [17, 19, 20, 22, 33, 34].

The effectiveness of randomized studies was evaluated using the RoB 2 probability of bias method. Most of the investigated studies showed substantial bias risk, which diminished trust in the stated outcomes. Although some research raised some doubt, domains examining bias from the randomization procedure and from preferential dissemination of outcomes generally indicated low probability of bias. The disclosure of missing empirical results and the suitability of outcome measures were the two main worries. High risk of bias was found in every study evaluated incorporating the RoB 2 technique for cluster randomized clinical trials. The scheduling of participant selection and enrollment in connection to the date of randomization, as well as any variations from the proposed intervention, have raised questions.

Several studies failed to record data for all variables and other problems included the presentation of incomplete outcome measures, incomplete assessment of outcomes, and selective discolosure of results. In the uncontrolled studies, there were few problems with sample selection, intervention categorisation, or variations from the planned intervention. Confidence in the presented results was, however, diminished by worries about the absence of data on possible confounding and bias resulting from outcome assessment in some trials. No research discussed how to deal with missing data.In other investigations, selective disclosure of the results seemed to be a problem.

Quantitative findings

In Tables 1 and 2 we show details about the interventions used in studies, their major both qualitative and as well as quantitative outcomes and also we showed which intervention was preferred.

Acceptability

The acceptance of various AI-enhanced therapies was evaluated through 3 controlled studies and 2 independent studies. Social robot acceptance was uneven and fluctuated according to individual robot and also their scenario of application (eg, clinical situation vs entertainment situation) [44, 45]. In two investigations, wearable sensor devices and environmental sensor devices were not well received, however one research received mixed reviews from LTC patients and carers [15]. Two of three carefully supervised studies examined social robotics. During a 12-week program, residents of nursing homes and hospitals expressed conflicting opinions of the robots like Guide and Cafero [14]. On the other hand, PARO was highly received and preferred to Guide in a dementia long term care facility [41]. An ecologically embedded mobility tracking system's poor acceptance among inhabitants, who doubted its utility, was revealed by 3rd controlled experiment [15].

Caretakers, who valued its simplicity of use and capacity to guarantee older patients' health, were more likely to find it acceptable. One of the two uncontrolled studies found that most participants found the social robotic MARIO (which attempted to lessen loneliness) comfortable, although there were flaws with look and availability as well as worries that persons with dementia would be cognizant of the robot's speech [30]. Another uncontrolled study examined a variety of health examining sensor devices that the participants found to be unsatisfactory and condemned [38] people have complained that the wearables sensor devices were cumbersome, the bed sensors kept them awake, and the continual observation bothered them. Other research covered outcomes including the development of bonds and decreased load on caretaker that might help make something more acceptable [13].

Depression symptoms

The usefulness of AI-enhanced robots in treating depression symptoms was examined in 6 controlled experiments. There were no appreciable variations seen between control category and intervention categories in five investigations. In a short trial involving dementia patients, Liang and colleagues discovered that depression symptoms had decreased from start of study in both the PARO category and the control category [18]. Despite the fact that this impact was only noticeable after 6 weeks psychological distress rose in the PARO subgroup whereas they did not in the comparison category after 12 weeks, . There were no appreciable differences in depressed symptoms as observed between the intervention subgroup and control sub-groups in two investigations assessing AI based environmental sensor devices [24, 25]. After evaluating baseline observations and follow-up results from five uncontrolled studies investigating robots with smart home scenarios, depressed symptoms showed some remission.

Quality of life

Only 2 of 5 experimental studies that looked at the impact of AI-improved robots on life quality revealed a meaningful difference. Both studies evaluated PARO's impact. Using a cross-over methodology, Moyle and colleagues exposed participants to PARO for duration of five weeks, followed by a three week rest interval. The standard of Life in patient suffering from Alzheimer’s Disease (QOL-AD) assessment was found to be moderately positively influenced by PARO [20]. In the second research, a genuine trained dog and PARO were put side by side in a skilled nursing facility [28]. Exposure happened twice a week for three months.

The Overall Quality of Life during Late-Stage Neurodegeneration (QUALID) scale ratings showed statistically significant variations, with the PARO subgroup scoring lower than the control sub-group on the measure. Three uncontrolled studies that looked at the effect of social robotic systems on standard of life came up with inconsistent findings, just one research suggesting a benefit.

Agitation

There were conflicting findings in five controlled trials that looked at the impact of pet robotics on agitation. While two trials failed to uncover any changes seen between control intervention categories, 3 studies discovered minor but significant reductions in restlessness [16, 18, 19, 21, 23].

Social outcomes

The impact of social robotics on loneliness was examined in two repeatable experiments and two uncontrolled trials. Controlled trials revealed that AIBO and PARO greatly reduce loneliness [13, 26]. In uncontrolled studies, it was demonstrated that PARO and NAO reduce loneliness [32, 34]. A total of five research evaluated social interaction. Communication between participants and nursing home staff improved significantly, according to a controlled trial employing PARO [18]. Following PARO therapies, interpersonal and interpersonal abilities significantly improved in two uncontrolled trials [35, 43]. Also, it was discovered that participants' feelings of social support were improved by the robot MARIO [33].

Behavioural outcomes

Individuals who communicated with PARO dramatically increased their general activity involvement and were more visually, physically and verbally engaged, according to 2 controlled studies and 1 non-controlled trial [21, 27, 43]. Studies discovered a reduction in negative behavior36 and an improvement in behavioural encouragement in 1 controlled study and 3 non-controlled studies that examined behavioural situations [40]. In contrast to some other research that demonstrated a significant decline in cognitive and behavioural scores in this cohort, a controlled trial indicated that introduction to PARO was linked to a higher prevalence of wandering (a disease-related behaviour that can make people disoriented or confused) in dementia patients [35]. There were three researches on level of dependency on others. After employing the Guide robotics or Cafero robots, participants' levels of reliance on others did not decrease much [14].

The extent to what environmental sensors devices can promote lack of dependency on others and influence daily life, however, is debatable [24, 39]. An environmentally implanted sensor system was the subject of a controlled research that revealed no significant variations in rate of hospitalisations [24]. However, a wearable sensor called CarePredict marketed by CarePredict, USA was related to reduced rates of hospitalisation.

Neuropsychiatric and cognitive outcomes

Cognitive performance was examined in five research. In 2 controlled studies, environmental sensor mechanisms have been shown to have no discernible impact on cognitive performance. 24,25] An independant trial, however, showed a notable increase in cognitive performance [37]. While some publications claim that PARO has no appreciable impact on cognition performance [28, 35] or neuropsychological symptoms18, others contend that PARO is linked to a reduction in nocturnal behavioural issues [28].

Results of physical capacity

PARO was used in two controlled experiments to measure muscle movements. While one study revealed no discernible effect [23], another discovered that it helped to decrease body movements [19]. Despite finding of no changes in balance, gait metrics and physical ability have been demonstrated to better in 2 controlled clinical trials using environmental sensing systems [24, 25]. Accidents of falls were not observed to be significantly affected by a worn sensor [29].

Psychological outcomes

Anxiety levels were discovered to be affected by social robots in different ways. Some studies have found no meaningful effect after exposure to PARO, while others have seen a slight reduction in anxiety [23]. NeCoRo had no discernible impact on anxiety [19]. Four experiments utilising PARO [20, 22, 31, 35] and one experiment using NAO demonstrated an improvement in emotional and mood states [34]. Having followed a PARO intervention, no appreciable effects in apathetic levels were seen [20]. However in a trial that used NAO, apathy considerably diminished [28].

Other health outcomes

PARO's impact on the sleep experience was examined in two controlled studies, the findings of which were inconsistent. While one research found a significant decrease in midday sleeping and an elevation in daytime awake, another found no evidence of improved sleep habits [22]. A wearable sensor device for the environment and an uncontrolled trial showed a considerable increase in duration of sleep [37]. Results for blood pressure as well as heart rate are inconsistent. There were no discernible variations in these consequences between the subjects in the PARO management category and the control category according to a protracted controlled experiment [18]. Conversely, a non-controlled research found that PARO was connected to a brief, significant reduction in heart rate and blood pressure [42]. There was no discernible impact of PARO on pain episode [23], cortisol [18].

Discussion

Although there was some disagreement on the acceptance of AI-enhanced treatments in LTC settings, this review indicated that there was little consensus about the efficacy of those initiatives for older individuals. Social robots have been shown to increase social interaction and mood, but the data was more conflicting and less definitive for the other innovations and consequences. The majority of research evaluated a variety of results, which made it impossible to synthesise them in a meaningful way and prevented a meta-analysis. In addition, many studies have moderate to severe bias risks due to underpowered design.

Contextualisation

For older adults undergoing LTC, this evaluation provides the first comprehensive look into the tolerability and efficacy of AI-enhanced therapies. The findings are consistent with earlier studies showing the acceptance and initial efficacy of AI-supplemented robotics for enhancing the mental health outcomes of senior citizens receiving LTC services [7, 9]. Relatively similar issues to those raised in this analysis have been identified in published studies of AI-supplemented intervention studies conducted in different health-care scenarios [46]. Improper reporting, limited sample numbers, a lack of outside confirmation of outcomes, problems with the concealment of result evaluations, and insufficient outcome information have been some of these [7, 9]. These analytical restrictions, nevertheless, might be explained by problems in the field. It is challenging to get a big sample because robots are expensive and scarce.

Furthermore, the absence of low-income and middle-income country coverage in research may be explained by the expensive cost of robotics. The fact that there are frequently technical dependability problems hinders the ability to repair the devices while they are being used in studies. Many robots are still in the prototype stage and lack the durability of widely used devices like mobile phones. Due to disease, exhaustion, and mortality, it is also challenging to collect follow-up information from older persons in LTC environments for self-report questionnaires. A more practical method of conducting fieldwork may be necessary for conducting research in the LTC context..

Implications

Ethical Implications

Ethics for AI-supplemented interventions that are offered as a component of LTC solutions have been the topic of numerous research projects [6, 47]. Elderly individuals who suffer from dementia may be led to believe the robotic is a genuine pet, [47], and some technologies may put them at risk of becoming infantilized. Some elderly persons develop bonds with robots and become upset when they are split up at the conclusion of a trial [48]. The level and recurrence of separation anxiety with robotics, however, are poorly understood, and it is still not obvious what approaches should be taken to end older people's connections with robotic systems. AI-supplemented sensors could lead to a loss of personalization in care and raise privacy concerns in LTC users [6]. In fact, numerous of the included research mentioned worries about monitoring and data privacy.

In certain trials, worries subsided with time as system trust increased. Future studies should pay attention to how well LTC customers, carers, and facility personnel are informed about privacy and security of information procedures [49]. The use of these innovations in LTC services and their acceptance depend critically on robust data privacy safeguards. Social robots can elicit a range of reactions from older individuals [6, 14]. While some elderly individuals have demonstrated positive engagement with robots, others, maybe as a result of higher baseline agitation, have showed apathy or unfavourable responses [14]. Yet, several research revealed that acceptability increased over time [7].

While adopting the robot, it is important to take into account some reports of protectiveness towards social human - robot interaction, particularly PARO. 6 These observations have practical consequences, demonstrating that to guard against unintended damages, individual desires must be taken into account as well as an implementation approach. Research analysing sensors have brought up questions about the scheduling of equipment placement, which calls for more thought in subsequent work. Studies involving dementia patients specifically show that placement should happen before more severe dementia begins to develop and worsen [49, 50].

Limitations

There were a number of restrictions on this review, that may have impacted the results. The inquiries were carried out using English search phrases in English-language resources, even though the google searches were not language-restricted. This may have restricted the foreign language outcomes to articles with keywords or abstracts in English language. Indeed, a number of results in other languages were discovered. Yet, by using interpreted search phrases in foreign language libraries, more publications could have been found. Also, a search of the computational scientific and technological literature was conducted using Web of Science, an integrative database that includes IEEE Xplore. Other machine learning databases may have been searched, possibly yielding more leads, albeit fewer likely clinical studies.

Future research and technology development

This evaluation offers guidance for future investigation and creation of therapies enhanced by AI in LTC services for geriatric patients. Several AI-enhanced interventions, such as conversational bots and smartphone apps, have not yet undergone clinical studies with senior citizens receiving LTC. It's likely that interventions utilising computers or smartphones will be more affordable and scalable ways to support LTC services. Indeed, Alexa from Amazon might help LTC patients maintain their independence, maintain their welfare, and connect them to tools and other people, albeit privacy issues would have to be resolved [51].

Future experiments should assess whether screen-based, AI supplemented interventions can offer distinct accessibility benefits over robots, even in settings with limited resources. Furthermore, given the fact that there have been no research investigations to date, AI-supplemented therapies for LTC facilities in low income countries and middle-income countries should be investigated. According to this analysis, AI-supplemented interventions did not properly cover the range of LTC services. The majority of interventions concentrated on psychological, behavioral, or psychosocial factors. The management of chronic health issues and helping older individuals with daily life tasks require more interventions.

In the interest of identifying outcomes including frailty, fractures, irregular lifestyle in dementia patients, disease relapses, slips, psychiatric illness, and alterations in medical status, the analysis discovered an array of environmental sensor system and wearable sensor system that integrated AI methodologies. The innovations, however, were still in the early stages of development and hadn't been independently evaluated for use in LTC services. The difficulties of conducting research in this area may be overcome by increasing financing for the subject and by fostering integrative partnerships between healthcare professionals and software professionals that uphold scientific credibility [52].

Conclusion

It is challenging to determine whether AI supplemented technologies for geriatric patients are significantly beneficial. Although some encouraging findings were made, more study is required. In order to make the available evidence more easily accessible, results for AI -supplemented intervention studies should be formalised, preferably with a triage exercise. They ought to be taken into consideration for scientific proof suggestion in the future because of the considerable variability of research design, number of respondents, the technological aspect of treatment outcomes, and reporting. Future research must be sure to provide the solutions that are best suited to meet the needs of this geriatric population receiving LTC, while also acknowledging that not all people will gain from these innovations in all circumstances. Until then, AI supplemented interventions might be viewed as a contest to develop new technologies rather than as viable options for geriatric patients involving LTC delivery.