Artificial intelligence-based clinical decision support systems (AI-CDSS) are emerging tools leveraging

machine learning and healthcare data to aid pharmacists in managing clinical complexity. They offer real-time

insights for identifying high-risk prescriptions, preventing drug interactions, and aiming to improve patient

outcomes. While the strength of AI-CDSS lies in data-driven support, human-centered design focusing on trust

and usability is crucial for adoption. Applications of AI-CDSS extend to patient care (e.g., disease prediction,

medication adherence) and pharmacy practice, including prioritizing prescription reviews (e.g., Lumio

Medication) and early detection of cognitive impairment. Although AI-CDSS shows potential in optimizing

pharmacy workflows and identifying drug-related issues, direct evidence of improved patient outcomes within

Today's pharmacists are central to patient safety, but managing increasing clinical complexity is no small feat.

That's where artificial intelligence-based clinical decision support systems (AI-CDSS) come in. These tools

combine machine learning with healthcare data to support pharmacists in identifying high-risk prescriptions,

preventing drug interactions, and improving patient outcomes. The strength of AI-CDSS lies in their ability to

deliver real-time, data-driven insights—empowering pharmacists to act faster and more confidently. However,

their success depends on more than just algorithms. A human-centered approach, emphasizing trust,

explainability, and ease of use, is crucial for adoption in real-world settings (1). Recent advancements in

medical sciences focusing on machine learning-based AI models, like AI-CDSS, help pharmacists prioritize

interventions by predicting which prescriptions may require attention. These tools are changing the way

human variables (such as explainability, privacy, and justice) as well as the social and organizational aspects of

medical practice play an equally significant role in deciding the success of AI-CDSS implementation, even in

spite of the technological improvement that has actually boosted AI-CDSS (2). AI-empowered CDSS

interventions (AI-CDSS) are anticipated to have a significant impact on the patient-provider relationship and

transform clinical decision-making (3).

Diabetes prediction using machine learning (ML)-CDSS and a comparison between traditional ML and deep

learning (DL) techniques. The results of their research indicated that random forest outperformed support

vector machines (SVM) and convolutional neural networks (CNN) (4). According to a study that used an ML

model to forecast tooth loss, models that take into account extra socioeconomic factors outperform those that

Neural Network (CNN) model was created in one study to detect tuberculosis infection from X-ray pictures.

They evaluated the suggested model using a data set of 4701 X-ray pictures, 453 of which were normal and

4249 of which were aberrant. The accuracy of the model they obtained was 85.68%. (8). Another study used a

CNN model to classify chest X-ray images, but instead of detecting only tuberculosis, it also found pleural

thickening, otosclerosis, and pulmonary interstitial hyperplasia. Over 16,000 annotated X-ray pictures were

needed to train this multi-classification model, which produced an accuracy of 82.2% (9). Data from low- and

mid-income countries are being used for AI research, according to a study that reviewed AI health applications

in various nations. To diagnose strokes and provide treatment plans, an expert system with a neural network

model and production rules was suggested (10).

study, hospital pharmacists use the Lumio Medication hybrid technology to calculate the likelihood that a

prescription may require PI. By classifying high- and low-risk medicines, the goal is to better prioritize

workflow and enhance patient outcomes (12). Prescriber behavior, data from electronic health records, and

pharmacy orders are all integrated into a machine learning categorization model. Hospital pharmacists are

expected to use this technology to identify and categorize prescription orders that need PI. Without

endangering patient privacy or data security, the goal was to continue improving patient health outcomes

through PI while lessening the workload of the pharmacists responsible for their care (13). Community

pharmacists can more effectively screen for people with early signs of MCI by using another ML data-driven

model that used risk factor data from the literature. This enables the pharmacist to refer patients who may have

early-stage MCI to the right medical services for diagnosis (14).

AI-CDSS can improve health care delivery in a number of ways. These include applications for population

health management (e.g., public health surveillance or predictive epidemiological modelling), diagnostic (e.g.,

early cancer diagnosis, diabetes retinopathy screening, or COVID-19 diagnosis based on computed

tomography images), therapeutic (e.g., precision medicine in chemotherapy and for combination drug therapy),

difficulties if the system generated an incorrect prediction. These would include the extra work of following up

on a referral, the expense of managing an incorrect diagnosis, and the psychological stress that a false positive

or other incorrect prediction could cause (19). Due to attitudinal hurdles, clinicians may be reluctant to

embrace or use AI-CDSS because they do not trust or comprehend its potential (20). The acceptance and

development of trust in AI-CDSS may also be hampered by a discrepancy between the information needs of

physicians and the outputs of the system (21). Clinicians find it difficult to understand how to use AI in CDSS,

and it takes up valuable time. The onboarding procedure and the absence of comprehensive training made

these problems worse (22).

Artificial intelligence-based clinical decision support systems (AI-CDSS) are emerging tools leveraging

pharmacy workflows and identifying drug-related issues, direct evidence of improved patient outcomes within

pharmacy remains limited. Implementation faces challenges like technical constraints and workflow

misalignment. This research synthesizes current research, highlighting the potential of AI-CDSS to transform

pharmacy practice while acknowledging the need for further investigation into their impact on patient

outcomes and the practical barriers to their widespread use.

1. Wang L, Zhang Z, Wang D, Cao W, Zhou X, Zhang P, Liu J, Fan X, Tian F. Human-centered design

and evaluation of AI-empowered clinical decision support systems: a systematic review. Frontiers in

Computer Science. 2023 Jun 2;5:1187299.

2. Magrabi F, Ammenwerth E, McNair JB, De Keizer NF, Hyppönen H, Nykänen P, Rigby M, Scott PJ,

machine learning and deep learning techniques. In2019 1st International informatics and software

engineering conference (UBMYK) 2019 Nov 6 (pp. 1-4).

5. IEEE.Elani HW, Batista AF, Thomson WM, Kawachi I, Chiavegatto Filho AD. Predictors of tooth

loss: A machine learning approach. PLoS One. 2021 Jun 18;16(6):e0252873.

6. Rosenfeld A, Benrimoh D, Armstrong C. Big data analytics and AI in mental healthcare,(2019). arXiv

preprint arXiv:1903.12071. 1903.

7. Frangou S, Sachpazidis I, Stassinakis A, Sakas G. Telemonitoring of medication adherence in patients

with schizophrenia. Telemedicine Journal & E-Health. 2005 Dec 1;11(6):675-83.

8. Liu C, Cao Y, Alcantara M, Liu B, Brunette M, Peinado J, Curioso W. TX-CNN: Detecting

tuberculosis in chest X-ray images using convolutional neural network. In2017 IEEE international

system for stroke diagnosis. InPhotonics Applications in Astronomy, Communications, Industry, and

High Energy Physics Experiments 2017 2017 Aug 7 (Vol. 10445, pp. 1060-1065). SPIE.

11. Corny J, Rajkumar A, Martin O, Dode X, Lajonchère JP, Billuart O, Bézie Y, Buronfosse A. A

machine learning–based clinical decision support system to identify prescriptions with a high risk of

medication error. Journal of the American Medical Informatics Association. 2020 Nov;27(11):1688-94.

12. Levivien C, Cavagna P, Grah A, Buronfosse A, Courseau R, Bézie Y, Corny J. Assessment of a hybrid

decision support system using machine learning with artificial intelligence to safely rule out

prescriptions from medication review in daily practice. International Journal of Clinical Pharmacy.

2022 Apr;44(2):459-65.

13. Balestra M, Chen J, Iturrate E, Aphinyanaphongs Y, Nov O. Predicting inpatient pharmacy order