Machine Learning-Based Clinical Decision Support Systems in

Dementia Care

Ritwik Raj Saxena

and Arshia Khan

Department of Computer Science, University of Minnesota, Duluth Campus, Duluth, Minnesota, U.S.A.

Keywords: Dementia, Clinical Decision Support Systems, Deep Learning, Caregiver Support.

Abstract: Clinical Decision Support Systems (CDSSs) enabled by machine learning (ML), particularly those based on

deep learning (DL), are revolutionizing dementia care by offering advanced capabilities that go beyond the

capacities of manual CDSSs, rule-based CDSSs, and statistical CDSSs. This paper explores unique

applications of ML in dementia care. It focuses on areas where ML-based, especially DL-based (neural

network-based) CDSSs currently excel and can potentially be of relevance. Unlike conventional CDSSs,

which evidently struggle with the complexities of large, heterogeneous datasets, ML models, particularly DL-

based ones, are capable of better identifying hidden patterns and subtle relationships across diverse genetic

and multi-omic, clinical, behavioural, socioeconomical and cultural, and environmental data. These systems

also extend their utility beyond clinical decision-making and caregiver wellbeing through tailored support

recommendations and aiding hospital administrators in resource mobilization, staff augmentation, and policy

formulation. However, challenges such as model interpretability, extensive data requirements, and

infrastructure limitations must be addressed. This article highlights the importance of a collaborative

approach, where various stakeholders in dementia come together to pool data and recommendations that

would assist in inculcating comprehensiveness and inclusivity in future CDSSs. We hypothesize that as DL

continues to showcase its decided prowess in the arena of decision-making, its applications in CDSSs will

keep playing an exceedingly pivotal role in advancing the efficacy of dementia care, improving patient

outcomes, and shaping the future of healthcare.

1 INTRODUCTION

In healthcare and medicine, decision support is crucial

for a variety of stakeholders, each of whom can utilize

customized assistance to eventually benefit those who

receive care. Decision support helps the stakeholders

optimize care delivery and patient outcomes. Such

stakeholders are many, but the primary ones are

healthcare providers and policymakers, whose verdicts

can have a momentous bearing on those who are on the

receiving end of the impact of their decisions. The

chief people involved in enriching their decisions with

evidence are data collectors, as well as researchers and

analysts. The entire pipeline or the process of

collecting data, quarrying and assembling trends,

patterns and insights from it, and using these analytics

to inform decision making is encapsulated in what is

called a decision support system (DSS).

https://orcid.org/0009-0001-7876-3193

https://orcid.org/0000-0001-8779-9617

DSSs are information-powered systems relying

on statistical and factual data to process it into

actionable knowledge, which is often in the form of

visualized data (a comprehensible, interest-piquing

format), and which fuels rational decision-making.

Although principally DSSs involve computational

components, there can be DSSs which do not involve

any computer-based tools or design. The idea of DSS

was conceptualized in 1960s (Keen & Morton, 1978).

When applied to clinical settings, DSSs are termed

Clinical Decision Support Systems (CDSSs), and

their major function here is aiding decision-makers in

medical diagnostics. CDSSs evolved through four

stages, standalone CDSSs, integrated systems,

standards-based systems, and service models which

(Wright & Sittig, 2008).

CDSSs can be typified on multiple bases. The first

basis is their function, or rather their style of function.

664

Saxena, R. R. and Khan, A.

Machine Learning-Based Clinical Decision Support Systems in Dementia Care.

DOI: 10.5220/0013194100003911

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 18th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2025) - Volume 2: HEALTHINF, pages 664-671

ISBN: 978-989-758-731-3; ISSN: 2184-4305

Some answer the question “what to choose”, while

others tell users “what to do.” Active and passive

CDSSs are other clades of CDSSs. Passive ones act

on a user-provided cue, while active ones provide

automated alerts and act on their own (Wasylewicz &

Scheepers-Hoeks, 2018).

While CDSSs that are completely manual are not

only possible to be but also to have existed, when

thinking about CDSSs currently, it is impossible to

find a person who would describe CDSSs without an

implication that these systems are run on, dependent

on, or directly associated with computers and are

computational tools. CDSSs are manifestations of

evidence-based medicine and therefore, rely on

statistical data and its analysis to serve as evidence for

its suggestions. The strongest tool presently available

in the realm of analysing data and drawing the most

meaningful insights from it is ML. A key area of AI,

ML automates detection, classification, prediction,

and generative tasks with high efficacy.

CDSSs can have a variety of applications in

medicine and healthcare. During the prescription

and medication process, the presence of CDSSs

ensures default values for drug doses and

frequencies and advise about patient-specific drug

allergies and adverse drug interactions. By reducing

manual transcription errors and ordering, a

computerized provider order entry (CPOE) system

enhanced with a CDSS mitigates treatment-

associated expenditures. CDSSs can reduce dosing

errors by 55% and improve adherence to evidence-

based protocols (Kuperman et al., 2007; Saxena &

Saxena, 2024). CDSSs considerably enhance

diagnostic accuracy. It offers symptom-specific

guidance and provides auto-analysis of complex

patient data, including medical history, test results,

and risk factors (Sutton et al., 2020), evidence-based

recommendations, and real-time support (Zhao et

al., 2023). Chronic conditions such as cancer,

chronic obstructive pulmonary disease, and

dementia require continuous monitoring and

evidence-based management, thereby benefiting

from use of CDSSs (Gencturk et al., 2024).

CDSSs improve healthcare delivery at

population level by tracking key performance

indicators (KPIs) and inspiring adherence to clinical

guidelines. CDSS-powered predictive analytics

enables outcome prediction (such as the likelihood

of hospital re-admission, disease progression) and

mortality and risk stratification of patients,

prioritizing interventions for those at higher risk

(Snooks et al., 2018). Natural language processing

(NLP) enables CDSSs to extract meaningful insights

by analysing unstructured data, such as clinical

notes, patient histories, and medical literature

(Berge et al., 2023). CDSSs power personalized

medicine by tailoring treatment and intervention

recommendations to individual patients based on

their genetic and multi-omic profile, environmental

factors, personal and familial history, and lifestyle

(Zhao et al., 2023).

CDSSs enable successful drug repurposing,

where new therapeutic uses for existing drugs are

identified (Zong et al., 2022). CDSSs also facilitate

real-time remote patient monitoring through

integration with telehealth services via linkages with

wearable devices, smart health apps, and home

monitoring. A CDSS can directly interact with

patients online (telemedicine) to provide virtual

consultation in the form of a text- or speech-based

intelligent conversation agent (Jadczyk et al., 2021),

ensuring patient empowerment.

CDSSs are applicable in toxicology management

by predicting patient poisoning and identifying its

cause of poisoning (Badcock, 2000). They assist in

the design and analysis of clinical trials (Embi et al.,

2009). For holistic care, CDSSs integrate herbal and

alternative therapies into its treatment

recommendations. CDSSs enable patient

engagement and patient confidence-building by

providing patients with personalized information

and education about their health conditions through

patient portals, and empowering them to access their

health records, to schedule appointments, and to

communicate with their healthcare providers

(Dendere et al., 2019; Hägglund et al., 2022).

Shared decision-making models, which use

CDSSs to present options based on patient-specific

data, improve patient satisfaction and adherence to

patient-opted treatment plans (Zhao et al., 2023;

Breitbart et al., 2020). The use of decision aids in

patient-centred care increases patient knowledge and

lowers decisional conflict (Alden et al., 2013).

CDSSs help hospital administrators by causing

improvement in operational efficiency through better

resource planning advice and more value-added

healthcare regulation adherence recommendations.

By using CDSSs, hospitals extenuate chances of

medical errors, whereby they save on litigative costs.

By avoiding unnecessary and duplicate tests and

medications as decision proxies, CDSSs save clinical

time and improve medicinal efficiency, through

which they diminish patient costs as well as clinical

costs (White et al., 2023).

Machine Learning-Based Clinical Decision Support Systems in Dementia Care

665

2 APPLICATIONS OF CDSSs IN

DEMENTIA CARE

CDSSs offer several applications that cater

specifically to the challenges posed by dementia,

which forms a class of neurodegenerative diseases

characterized by cognitive and motor decline and

include Alzheimer’s disease as well as Lewy Body

dementia (Bruun et al., 2019). Early detection of

dementia helps slow down the progression of the

ultimately mortal disease (Rasmussen & Langerman,

2019). CDSSs analyse data such as neuroimaging,

genomic and multi-omic data and genetic markers,

individual symptoms, health status, past conditions,

as well as family history of a suspected person with

dementia (PwD), their cognitive test results, and

pathological test results, if relevant, to identify

dementia as well as prodromal dementia (Kleiman et

al., 2022; Karimi et al., 2020; Saxena et al., 2023;

Johnson et al., 2020) long before noticeable

symptoms arise. Timely intervention greatly

enhances the quality of life for patients and can arrest

the acceleration of evolution of dementia.

CDSSs engender personalized cognitive care

planning in dementia care. CDSSs assess cognitive

decline trajectory of PwDs and recommend tailored

interventions, inter alia, memory exercises, serious

games, lifestyle changes including physical exercise

and dietary changes, pharmacological and

physiotherapeutic treatments based on the

individual’s history, stage of dementia, mobility

status, and progression patterns (Dietlein et al., 2018).

A paradigm of personalized care is immensely

applicable to dementia because in this disease,

gender-, population-, age-, region- and patient-level

specificity is observed not only with regards to

symptomatology but also with regards to the pattern

and speed of progression.

No two PwDs can be, with arrant certainty, said to

display the same behaviour (behavioural variability)

given the similarity in their ages, stage of dementia,

level of cognitive decline, gender identicality and

overlap of other factors. Genetic markers, individual

medical and psycho-emotional history, environment,

socio-familial dynamics and other factors influence

the behaviour of a PwD (Cerejeira et al., 2012).

CDSSs assist in management of behavioural, social,

and psychological symptoms like agitation,

aggression, depression and hopelessness, suicidality,

misanthropy, anxiety, cynicism, and other similar

issues in PwDs, specifically those in advanced stages

of dementia (Müller-Spahn, 2003), in identifying

triggers and recommending treatments.

Caregiver support directly influences and

augments patient care. CDSSs play a critical role in

it. Caring for PwDs is emotionally and physically

taxing for their caregivers, whether the caregivers are

hired or are informal caregivers (Lin et al., 2023).

CDSSs provide caregivers with real-time guidance on

managing challenging behaviours, alert them to

critical changes in the PwD’s condition, and offer

widely applicable educational resources linked to

dementia and other health conditions as well as those

resources which are tailored to a PwD’s specific stage

of dementia. CDSSs also enable a caregiver to relay

real-time alerts and warnings to healthcare

professionals in case their ward, faces a medical

emergency (Ponnala et al., 2020).

Caregivers, both formal and informal, are key

users of CDSSs, which are their vital aides. CDSSs

provide them with behaviour management tools and

educational aids that provide guidance on handling

the emotional and behavioural symptoms of their

wards, as well as other pointers about suitable

caregiving (Duplantier & Williamson, 2023; Antelo

Ameijeiras & Espinosa, 2023). CDSSs also link them

to resources which can uplift them when they feel

overwhelmed by their duties or by perceived role

captivity. CDSSs provide caregivers with resources

on disease progression, care techniques, and

strategies for managing activities of daily living

(ADLs). For caregivers in general, and particularly

for family (informal) caregivers, who juggle other

responsibilities, CDSSs act as digital assistants who

remind them about their wards’ appointments and

medication and care schedules.

Longitudinal tracking of cognitive decline is a

significant feature of CDSSs in dementia (Rebok et

al., 1990). CDSSs, using sensor data, IoT, and

caregiver inputs, monitor changes in cognitive

function of PwDs over time. CDSSs suggest

reassessment in interventions as needed.

CDSSs in dementia care are integrated with

network-enabled assistive technologies (Internet of

Things or IoT), such as wearable devices like

smartwatches, smartphones, intelligent exoskeletons

(for those dementia patients with mobility

impairments) and posture and balance assessment

devices, home monitoring systems in smart homes,

car dashboard assistants (for patient), virtual reality

devices, hearing devices and aids, and so on, to gather

data, inform healthcare providers and caregivers of

real-time updates about PwDs’ condition, and to

support the patients in routine tasks and ADLs. These

systems monitor vital signs, detect falls, and even

assess whether a patient is adhering to prescribed

daily routines. This underpins continuous and

HEALTHINF 2025 - 18th International Conference on Health Informatics

666

unobtrusive tracking of patients’ symptoms and

information connected to their adherence to treatment

plans.

For physicians, nurses, and allied healthcare

professionals engaged in treating PwDs, CDSSs

assist in diagnosis, personalized treatment planning,

and medication management (for example, for

dementia patients who take multiple medications for

comorbid conditions), real-time clinical guidance,

identifying trends in patient incidents (e.g., falls,

medication errors), implementing preventive

measures to improve patient safety, and protocols for

managing behaviour in PwDs (Lindgren, 2011).

CDSSs improve operational efficiency and patient

flow, optimize resource allocation (by demonstrating

accuracy in predicting patient needs) and staffing

requirements in dementia care units in hospitals

(Chen et al., 2023).

For administrative purposes, CDSSs provide

quality assurance services such as tracking clinical

outcomes, clinical compliance with dementia care

guidelines, and collecting data and providing insights

by processing it and allowing its analysis to be used

for internal audits, accreditation processes, and

quality improvement initiatives. Dementia care unit

managers use CDSSs for eliciting actionable

recommendations for organizational risk

management.

For clinical investigators engaged in researching

dementia, CDSSs facilitate data collection and

analysis for ongoing studies. These systems are fed

large volumes of data from sources like electronic

health records (EHRs), conversation logs between

dementia stakeholders, particularly those between

PwDs and their caregivers or their healthcare

providers, clinical notes, patient reports, and

caregiver feedback. The analytics provided by CDSSs

based on such data empowers researchers to explore

new hypotheses about dementia’s pathology,

aetiology, progression, treatment and so on. CDSSs

help identify balanced patient cohorts (by applying

probability distributions and sampling techniques to

select highly representative groups as functional

samples) for clinical trials based on specific

biomarkers, symptoms, age, stage of disease, and so

on (Kwan et al., 2020). CDSSs perform predictive

analytics on dementia statistics and model dementia

patterns under varied intervention and socio-

environmental scenarios. This helps CDSSs discern

trends that are not apparent via manual analysis

(Gomez-Cabello et al., 2024).

Technologists and computer scientists are

involved in building and refining software that

powers CDSSs. They use the outcomes that

researchers generate, the analytics that CDSSs

generate, and the recommendations that other

stakeholders in dementia care (SDCs) tender to

inform their requirement gathering efforts for

improving extant CDSSs and creating new and better

ones (Elhaddad & Hamam, 2024).

3 MACHINE LEARNING-BASED

CLINICAL DECISION

SUPPORT SYSTEMS IN

DEMENTIA CARE

The current applications of AI-based or ML-based

CDSSs in dementia care are, straightforwardly or with

some level of indirectness, prodigiously concentrated

in the area of disease diagnostics in patients, that is,

determination, prediction, detection, classification,

recognition, or identification of, as the case may be,

the presence of dementia or cognitive decline in a

person, including mild cognitive decline as being

prodromal to dementia or not, and the severity or the

stage of dementia or cognitive decline as evident in a

PwD (Rhodius-Meester et al., 2018). Automating

diagnostics is a complex task which requires

optimization of a multivariate setup, which involves

numerous categorical and numerical variables.

Therefore, while some of the other decision support

tasks applicable to dementia care can be tolerably and,

sometimes, quite capably performed though other

techniques, diagnostics is best accomplished using DL

methods, particularly because DL models can handle

the non-linear relationships inherent in this task

(Javeed et al., 2023).

DL models analyze enormous datasets and learn

relevant features from raw data on their own. Their

latter capability remedies the need for deriving

features from the data manually or statistically. This

is remarkably useful in domains such as dementia

care where data is commonly dense and high-

dimensional. DL models have demonstrated state-of-

the-art performance on a variety of tasks, inter alia,

medical image analysis, NLP (implementation of

intelligent conversation agents in dementia care and

so on) and predictive modeling (Karako et al., 2023).

This makes them well-suited for determining

diagnosis and prognosis in dementia care.

Techniques have been developed so that DL

models can be expeditiously and unobtrusively

integrated with other existing components of a CDSS,

such as patient information systems (Maleki

Varnosfaderani & Forouzanfar, 2024). While it is also

indubitably true that diagnostics, being the principal

Machine Learning-Based Clinical Decision Support Systems in Dementia Care

667

manifestation of predictive analysis in medicine, and

predictive analysis being the mainstay of ML –

specifically DL – must be the primary use case of ML

in a CDSS, there are multiple other uses that DL can

perform when amalgamated in a CDSS.

One application of DL-enabled CDSSs in

dementia care is predicting the incidence of dementia

in a population based on large-scale datasets,

including genetic makeup of the population, regional

and environmental factors affecting the population,

lifestyle and cultural-anthropological characteristics

of the population, and historical dementia data for the

population (Kim & Lim, 2021). This can be done in

conjunction with studying how the incidence of

dementia in that population evolved over time

predicting, using DL, future trends of dementia

incidence in that population. This helps with better

resource allocation for the future dementia care for

that population, educating the population about

dementia and how to prevent it from affecting or

delaying when it is inevitable, and implementing

interceptive strategies that can prolong the onset and

potentially reduce the incidence of dementia in the

population.

Another critical application of DL is the analysis

of a mishmash of unstructured data (which not only

dementia, but also other domains of medicine are

replete with), including heterogenous and multimodal

data, first detecting where each piece of the disparate

data fits (assigning labels to it using unsupervised

learning) and then using those labels to extract

features from it, visualize it and make predictions and

extrapolations based on it (Taye, 2023). This requires

not only the NLP capability of ML but also its

competence in processing large quantities of

multimodal data.

DL is more adept at drug repurposing than

statistical CDSSs struggle. This is because DL-based

CDSSs can mine or are fed with humongous (at times,

almost exhaustive) amounts of drug-drug interaction

and molecular reaction data, which is commonly an

instance of high-dimensional data. Also, drug

discovery and drug-drug interactions are modeled far

more accurately by using DL than with statistical

CDSSs (Chen et al., 2024). This integrates data from

pharmacogenomics, multi-omics, clinical trial

outcomes, and patient histories, and helps prevent

adverse drug reactions and the possible occurrence of

toxicity in patients. ML enables personalized drug

therapy by considering individual patient data, such

as genetic profiles and disease progression patterns.

DL models can continuously analyze complex

datasets. They can provide personalized predictions

that adapt as new data becomes available, such as

changes in patient cognition and behavior (Arya et al.,

2023).

DL provides substantial benefits in decision-

making processes in areas of dementia where non-

linear data patterns need to be analyzed. While

statistical techniques display the ability to determine

the best intervention, including drugs and treatment

plans, in individual cases of dementia, this task is

performed far more accurately with DL models,

which they accomplish by analyzing astronomical

amounts of patient-specific data, including medical

history, genetic information, and real-time health

monitoring. Unlike traditional methods, DL identifies

subtler patterns in disease progression data compared

to statistical tools and suggests highly personalized

treatments (Alowais et al., 2023).

DL techniques also exhibit much better promise

in predicting the most appropriate approach to

caregiver wellbeing by analyzing psychological and

physical health data from caregivers. DL also informs

and enhances the design of the best administrative

policies in dementia care at the policy formulation

echelons, reducing staff burnout by optimizing shifts,

schedules, and employee numbers, concomitant with

dementia incidence. DL-based CDSSs personalize

cognitive exercises like serious games with

unparalleled precision (Maggio et al., 2023).

4 DISCUSSION AND

CHALLENGES

Implementing DL models, for CDSSs, presents

various challenges. The complexity and

interpretability of these models is a concern. Large

Language Models (LLMs) are NLP models which are

especially massive, and their inner workings are hard

to decipher.

DL algorithms are often described as “black box”

models due to the difficulty in understanding how

they arrive at certain decisions (Hassija et al., 2023).

In healthcare, where trust in technology is paramount,

the inability to explain a model's predictions hinders

its adoption (Rudin, 2019; Abgrall et al., 2024). To

combat this, explainable AI (XAI) techniques like

LIME and SHAP can be applied.

Another major challenge is the extensive training

time required by DL models to achieve satisfactory

performance metrics such as accuracy, precision, and

F1 score. These models often require substantial

computational power and vast amounts of data to

reach optimal results. In dementia care, relevant

clinical data might be scattered across multiple

HEALTHINF 2025 - 18th International Conference on Health Informatics

668

institutions and databases. Without sufficient data,

training DL models becomes less feasible.

The deployment and integration of DL-based

CDSSs come with infrastructural challenges. Smaller

healthcare facilities do not have the necessary

infrastructure to host DL models locally (Miotto et

al., 2018). Cloud-based solutions offer scalability in

such scenarios. However, they introduce concerns

about data privacy, security, and cost (Mehrtak et al.,

2021). They need regular updates and maintenance.

This is burdensome for underfunded healthcare

systems. Thus, infrastructural and technological

investments are crucial.

Healthcare providers in dementia care should

contribute data to a unified database. Extant

repositories like Alzheimer's Disease Neuroimaging

Initiative (ADNI) (Weiner et al., 2013) must be linked

to form a centralized registry. Healthcare providers

continuously generate new data. Insights from this

data could be incorporated into the CDSSs.

One approach for reducing the training and

running time of ML models is to ensure that the data

contributed to CDSSs is properly structured and

encoded so that ML models can interpret it

effectively. Structured data can be easily converted

into knowledge graph embeddings, which represent

complex relationships between different data points,

such as linking sociocultural information to clinical

symptoms and treatment outcomes. These

embeddings enable the ML models to process and

learn from the data more efficiently.

5 FUTURE DIRECTIONS AND

CONCLUSION

Our paper showcases that use of ML, particularly DL,

as the underlying force that fuels CDSSs in dementia

care can indeed serve significant improvements in

their ability to accrue benefits to PwDs, to other

SDCs, and to dementia care institutions, and

incentivize computer scientists and technologists in

this field. From early detection and diagnosis to

personalized treatment planning and monitoring, DL-

powered CDSSs have the potential to vastly improve

outcomes for patients with dementia and reduce the

burden on the healthcare provision and policymaking

superstructure. While challenges exist, they can

be overcome with conscientious research

and development efforts and technological

improvements. As research and development in this

area continue to advance, we can expect to see even

more innovative and impactful applications of DL in

dementia care. Collaboration between various SDCs,

while laying a particular emphasis on the involvement

of PwDs, researchers, experts, and technologists, can

help gather specific stakeholder needs, allowing for

the design and development of inclusive CDSSs

based on those needs. By working together with

different SDCs, we can access a wider range of

perspectives and details, leading to a more

comprehensive understanding of the necessary

information. Ensuring data privacy will remain key to

the success of CDSSs.

REFERENCES

Abgrall, G., Holder, A. L., Chelly Dagdia, Z., & Monnet,

X. (2024). Should AI models be explainable to

clinicians? Critical Care, 28(1), 301. https://doi.

org/10.1186/s13054-024-05005-y.

Alden, D. L., Friend, J., & Chun, M. B. (2013). Shared

decision making and patient decision aids: Knowledge,

attitudes, and practices among Hawai‘i physicians.

*Hawai’i Journal of Medicine & Public Health,

72*(11), 396. Available: https://pmc.ncbi.nlm.

nih.gov/articles/PMC3831568/.

Alowais, S. A., Alghamdi, S. S., Alsuhebany, N.,

Alqahtani, T., Alshaya, A. I., Almohareb, S. N., ... & Al

Harbi, S. (2023). Revolutionizing healthcare: The role

of artificial intelligence in clinical practice. BMC

Medical Education, 23(1), 689. https://doi.org/10.

1186/s12909-023-04698-z.

Antelo Ameijeiras, P., & Espinosa, P. (2023). The influence

of different types of social support in caregivers of

people with dementia. Int J Psychol Res (Medellin),

15(2), 85–93. https://doi.org/10.21500/20112084.5639.

Arya, A. D., Verma, S. S., Chakarabarti, P., Chakrabarti, T.,

Elngar, A. A., Kamali, A. M., & Nami, M. (2023). A

systematic review on machine learning and deep

learning techniques in the effective diagnosis of

Alzheimer's disease. Brain Informatics, 10(1), 17.

https://doi.org/10.1186/s40708-023-00195-7.

Badcock, N. R. (2000). Detection of poisoning by

substances other than drugs: A neglected art. *Annals

of Clinical Biochemistry, 37*(2), 146–157.

https://doi.org/10.1258/0004563001899140.

Berge, G. T., Granmo, O. C., Tveit, T. O., Munkvold, B. E.,

Ruthjersen, A. L., & Sharma, J. (2023). Machine

learning-driven clinical decision support system for

concept-based searching: A field trial in a Norwegian

hospital. *BMC Medical Informatics and Decision

Making, 23*(1). https://doi.org/10.1186/s12911-023-

02101-x.

Block, L., Lingg, A. J., Kind, A. J., & Werner, N. E. (2020).

Conceptualizing caregiving activities for persons with

dementia (PwD) through a patient work lens. Appl

Ergon, 85, 103070. https://doi.org/10.1016/j.apergo.

2020.103070.

Machine Learning-Based Clinical Decision Support Systems in Dementia Care

669

Breitbart, E. W., et al. (2020). Improved patient satisfaction

and diagnostic accuracy in skin diseases with a visual

clinical decision support system—a feasibility study

with general practitioners. *PLOS ONE, 15*(7),

e0235410. https://doi.org/10.1371/journal.pone.

0235410.

Bruun, M., et al. (2019). Impact of a clinical decision support

tool on dementia diagnostics in memory clinics: The

PredictND validation study. *Current Alzheimer

Research, 16*(2), 91–101. https://doi.org/10.2174/

1567205016666190103152425.

Cerejeira, J., Lagarto, L., & Mukaetova-Ladinska, E. B.

(2012). Behavioral and psychological symptoms of

dementia. *Front Neurol, 3*, 73.

https://doi.org/10.3389/fneur.2012.00073.

Chen, C. Y., et al. (2024). Ability of machine-learning

based clinical decision support system to reduce alert

fatigue, wrong-drug errors, and alert users about look

alike, sound alike medication. Computer Methods and

Programs in Biomedicine, 243, 107869. https://

doi.org/10.1016/j.cmpb.2023.107869.

Chen, Z., et al. (2023). Harnessing the power of clinical

decision support systems: Challenges and

opportunities. Open Heart, 10(2), e002432. https://

doi.org/10.1136/openhrt-2023-002432.

Dietlein, C., Eichberg, S., Fleiner, T., & Zijlstra, W. (2018).

Feasibility and effects of serious games for people with

dementia: A systematic review and recommendations

for future research. *Gerontechnology, 17*(1), 1–17.

https://doi.org/10.4017/gt.2018.17.1.001.00.

Dendere, R., Slade, C., Burton-Jones, A., Sullivan, C.,

Staib, A., & Janda, M. (2019). Patient portals

facilitating engagement with inpatient electronic

medical records: A systematic review. *Journal of

Medical Internet Research, 21*(4). https://doi.org/10.

2196/12779.

Duplantier, S. C., & Williamson, F. A. (2023). Barriers and

facilitators of health and well-being in informal

caregivers of dementia patients: A qualitative study. Int

J Environ Res Public Health, 20(5), 4328.

https://doi.org/10.3390/ijerph20054328.

Elhaddad, M., & Hamam, S. (2024). AI-driven clinical

decision support systems: An ongoing pursuit of

potential. Cureus, 16(4), e57728. https://doi.org/10.

7759/cureus.57728.

Embi, P. J., Kaufman, S. E., & Payne, P. R. O.

(2009). Biomedical informatics and outcomes

research: Enabling knowledge-driven healthcare.

*Circulation, 120*(23), 2393. https://doi.org/10.1161/

CIRCULATIONAHA.108.795526.

Gencturk, M., et al. (2024). Transforming evidence-based

clinical guidelines into implementable clinical decision

support services: The CAREPATH study for

multimorbidity management. *Frontiers in Medicine,

11*. https://doi.org/10.3389/fmed.2024.1386689.

Gomez-Cabello, C. A., Borna, S., Pressman, S., Haider, S.

A., Haider, C. R., & Forte, A. J. (2024). Artificial-

intelligence-based clinical decision support systems in

primary care: A scoping review of current clinical

implementations. European Journal of Investigation in

Health, Psychology and Education, 14(3), 685–698.

https://doi.org/10.3390/ejihpe14030045.

Hägglund, M., McMillan, B., Whittaker, R., & Blease, C.

(2022). Patient empowerment through online access to

health records. *BMJ, 378*, e071531. https://doi.

org/10.1136/bmj-2022-071531.

Hassija, V., Chamola, V., Mahapatra, A., Singal, A., Goel,

D., Huang, K., ... & Hussain, A. (2023). Interpreting

black-box models: A review on explainable artificial

intelligence. Cognitive Computation. https://doi.org/

10.1007/s12559-023-10179-8.

Jadczyk, T., Wojakowski, W., Tendera, M., Henry, T. D.,

Egnaczyk, G., & Shreenivas, S. (2021). Artificial

intelligence can improve patient management at the

time of a pandemic: The role of voice technology.

*Journal of Medical Internet Research, 23*(5), e22959.

https://doi.org/10.2196/22959.

Javeed, A., et al. (2023). Machine learning for dementia

prediction: A systematic review and future research

directions. Journal of Medical Systems, 47(1), 17.

https://doi.org/10.1007/s10916-023-01906-7.

Johnson, K. B., et al. (2020). Precision medicine, AI, and

the future of personalized health care. *Clinical and

Translational Science, 14*(1).

Karako, K., Song, P., & Chen, Y. (2023). Recent deep

learning models for dementia as point-of-care testing:

Potential for early detection. Intractable Rare Diseases

Research, 12(1), 1–4. https://doi.org/10.5582/irdr.

2023.01015.

Karimi, S., Jiang, X., Dolin, R. H., Kim, M., & Boxwala,

A. (2020). A secure system for genomics clinical

decision support. *Journal of Biomedical Informatics,

112*, 103602. https://doi.org/10.1016/j.jbi.2020.

103602.

Keen, P. G. W., & Morton, M. S. S. (1978). Decision

support systems: An organizational perspective.

Dssresources.com.

https://dssresources.com/books/contents/keen78.html

Kim, J., & Lim, J. (2021). A deep neural network-based

method for prediction of dementia using big data.

International Journal of Environmental Research and

Public Health, 18(10), 5386. https://doi.org/10.

3390/ijerph18105386.

Kleiman, M. J., Ariko, T., & Galvin, J. E. (2022).

Hierarchical two-stage cost-sensitive clinical decision

support system for screening prodromal Alzheimer’s

disease and related dementias. *Journal of Alzheimer’s

Disease, 91*(2), 895–909. https://doi.org/10.3233/jad-

220891.

Kuperman, G. J., et al. (2007). Medication-related clinical

decision support in computerized provider order entry

systems: A review. *Journal of the American Medical

Informatics Association, 14*(1), 29–40. https://doi.

org/10.1197/jamia.m2170.

Kwan, J. L., et al. (2020). Computerised clinical decision

support systems and absolute improvements in care:

Meta-analysis of controlled clinical trials. BMJ, 370,

m3216. https://doi.org/10.1136/bmj.m3216.

Lin, X., Moxley, J. H., & Czaja, S. J. (2023). Caring for

dementia caregivers: Psychosocial factors related to

HEALTHINF 2025 - 18th International Conference on Health Informatics

670

engagement in self-care activities. *Behav Sci (Basel),

13*(10), 851. https://doi.org/10.3390/bs13100851.

Lindgren, H. (2011). Towards personalized decision

support in the dementia domain based on clinical

practice guidelines. User Model. User-Adapt. Interact.,

21, 377–406. https://doi.org/10.1007/s11257-010-

9090-4.

Maleki Varnosfaderani, S., & Forouzanfar, M. (2024). The

role of AI in hospitals and clinics: Transforming

healthcare in the 21st century. Bioengineering, 11(4),

337. https://doi.org/10.3390/bioengineering11040337.

Maggio, M. G., De Bartolo, D., Calabrò, R. S., Ciancarelli,

I., Cerasa, A., Tonin, P., ... & Iosa, M. (2023).

Computer-assisted cognitive rehabilitation in

neurological patients: State-of-art and future

perspectives. Frontiers in Neurology, 14, 1255319.

https://doi.org/10.3389/fneur.2023.1255319.

Mehrtak, M., SeyedAlinaghi, S., MohsseniPour, M., Noori,

T., Karimi, A., Shamsabadi, A., ... & Mehraeen, E.

(2021). Security challenges and solutions using

healthcare cloud computing. Journal of Medicine and

Life, 14(4), 448-461. https://doi.org/10.25122/jml-

2021-0100.

Miotto, R., Wang, F., Wang, S., Jiang, X., & Dudley, J. T.

(2018). Deep learning for healthcare: Review,

opportunities and challenges. Briefings in

Bioinformatics, 19(6), 1236-1246. https://doi.org/10.

1093/bib/bbx044.

Müller-Spahn, F. (2003). Behavioral disturbances in

dementia. *Dialogues Clin Neurosci, 5*(1), 49–59.

https://doi.org/10.31887/DCNS.2003.5.1/fmuellerspahn.

Ponnala, S., Block, L., Lingg, A. J., Kind, A. J., & Werner,

N. E. (2020). Conceptualizing caregiving activities for

persons with dementia (PwD) through a patient work

lens. Appl Ergon, 85, 103070. https://doi.org/10.

1016/j.apergo.2020.103070.

Rasmussen, J., & Langerman, H. (2019). Alzheimer’s

disease—Why we need early diagnosis. *Degenerative

Neurological and Neuromuscular Disease, 9*(9), 123–

130. https://doi.org/10.2147/dnnd.s228939.

Rebok, G., Brandt, J., & Folstein, M. (1990). Longitudinal

cognitive decline in patients with Alzheimer's disease.

J Geriatr Psychiatry Neurol, 3(2), 91–97.

https://doi.org/10.1177/089198879000300207.

Rhodius-Meester, H. F. M., et al. (2018). Computer-

assisted prediction of clinical progression in the earliest

stages of AD. Alzheimer’s and Dementia: Diagnosis,

Assessment and Disease Monitoring, 10, 726–736.

https://doi.org/10.1016/j.dadm.2018.09.001.

Rudin, C. (2019). Stop explaining black box machine

learning models for high stakes decisions and use

interpretable models instead. Nature Machine

Intelligence, 1(5), 206-215. https://doi.org/10.1038/

s42256-019-0048-x.

Saxena, R. R., & Saxena, R. (2024). Applying graph neural

networks in pharmacology. *Authorea Preprints,

Techrxiv*. https://doi.org/10.36227/techrxiv.

170906927.71541956/v2.

Saxena, R., Carnevale, K., Yakymovych, O., Salzle, M., &

Sharma, K. (2023). Precision, personalization, and

progress: Traditional and adaptive assessment in

undergraduate medical education. *Innovative

Research Thoughts, 9*(4), 216–223. https://doi.org/10.

36676/irt.2023-v9i4-029.

Snooks, H., et al. (2018). Introduction – Predictive risk

stratification model: A randomised stepped-wedge trial

in primary care (PRISMATIC). *NIHR Journals

Library*. Available: https://www.ncbi.nlm.nih.gov/

books/NBK475995/.

Sutton, R., Pincock, D., Baumgart, D., Sadowski, D.,

Fedorak, R., & Kroeker, K. (2020). An overview of

clinical decision support systems: Benefits, risks, and

strategies for success. *NPJ Digital Medicine, 3*(1), 1–

10. https://doi.org/10.1038/s41746-020-0221-y.

Taye, M. M. (2023). Understanding of machine learning

with deep learning: Architectures, workflow,

applications and future directions. Computers, 12(5),

91. https://doi.org/10.3390/computers12050091.

Wasylewicz, A. T. M., & Scheepers-Hoeks, A. M. J. W.

(2018). Clinical decision support systems. *PubMed*.

https://www.ncbi.nlm.nih.gov/books/NBK543516/.

Weiner, M. W., et al. (2013). The Alzheimer’s Disease

Neuroimaging Initiative: A review of papers published

since its inception. Alzheimer’s & Dementia, 9(5),

e111–e194. https://doi.org/10.1016/j.jalz.2013.05.

1769.

White, N., et al. (2023). Evaluating the costs and

consequences of computerized clinical decision support

systems in hospitals: A scoping review and

recommendations for future practice. *Journal of the

American Medical Informatics Association, 30*(6),

1205–1218. https://doi.org/10.1093/jamia/ocad040.

Wright, A., & Sittig, D. F. (2008). A framework and model

for evaluating clinical decision support architectures.

*Journal of Biomedical Informatics, 41*(6), 982–990.

https://doi.org/10.1016/j.jbi.2008.03.009.

Zhao, C., et al. (2023). Harnessing the power of clinical

decision support systems: Challenges and

opportunities. *Open Heart, 10*(2), 1–11. https://

doi.org/10.1136/openhrt-2023-002432.

Zong, N., et al. (2022). Computational drug repurposing

based on electronic health records: A scoping review.

*NPJ Digital Medicine, 5*(1). https://doi.org/10.

1038/s41746-022-00617-6.

Machine Learning-Based Clinical Decision Support Systems in Dementia Care

671