© 2022 MJH Life Sciences and Pharmacy Times – Pharmacy Practice News and Expert Insights. All rights reserved.
© 2022 MJH Life Sciences™ , Pharmacy Times – Pharmacy Practice News and Expert Insights. All rights reserved.
By interpreting and applying the data provided by algorithms, pharmacists are uniquely equipped to guide development of more patient-centered care.
The definition of artificial intelligence (AI) continues to evolve but can be thought of as a simulation of human intelligence, specifically in terms of the way humans act and think. Machine learning (ML) is a technique of AI that involves training a model (or algorithm) to make predictions based on data input.1 AI has multiple potential uses in the clinical and industry settings, and some interesting real-world cases help illustrate these opportunities. With the advent of these technologies, however, there is also a need for further education on AI in health care curricula.
The digital health industry is growing at an extraordinary pace and has undoubtedly been accelerated by the COVID-19 pandemic. “Digital health” and “artificial intelligence” have become buzzwords both within and outside of health care. As we begin to see more of these technologies implemented in health care settings, it is essential for pharmacists to be involved.
Pharmacists are frontline health care professionals who have consistent touchpoints with patients. With digital health becoming a core component of chronic disease management, whether it be in symptom monitoring, adherence tracking, or even delivery of therapy itself, pharmacists play an integral role in leveraging technologies to improve patient experiences and outcomes.2 Pharmacists’ voices are needed at the table when discussing the development and implementation of AI algorithms and the need to advocate for the integration of a more robust AI component in the pharmacy curriculum. Pharmacists can also encourage students and other practicing pharmacists to be curious about utilizing AI in their practice.
As medication experts, pharmacists help to identify optimal treatment for patients across hospital, primary care, and community settings. Pharmacists undergo rigorous training to develop a critical thinking framework for patient work-up. With much of digital health technologies involving an element of AI, pharmacists will need to apply the same skillsets to evaluate the underlying algorithms. Although many clinical applications of AI are still in the proof-of-concept stage, there seems to be a lack of urgency in establishing pharmacists as key stakeholders. This article will share an overview of existing applications of AI in the context of pharmacy and discuss opportunities for involvement by pharmacists.
Pharmacists in various health care settings have used AI to provide data-driven interventions via Clinical Decision Support Systems (CDSS). This type of technology helps pharmacists comb through data and make interventions to prevent medication errors, minimize patient complications, and save costs.3 For example, one company, Arine, has created a platform that provides pharmacists with necessary data to deliver comprehensive patient care services such as tailored medication management, lifestyle counseling, and care coordination through telehealth.4
Another company, Cricket Health, utilizes ML to support patients with chronic kidney disease (CKD). The Cricket Health models are designed to predict estimated glomerular filtration (eGFR) without lab data. A cohort study was designed using Cricket Health’s educational program with 37 participants in the intervention arm (61 total). Survey results revealed that more patients in the Cricket Health program started dialysis in the outpatient setting and demonstrated better knowledge of the disease (p<0.001).5
A separate, similar study with Cricket Health involved an educational program for patients with CKD stages 4 and 5 to help determine which treatment modality would be best for their renal health. The patients showed increases in peritoneal dialysis selection, renal transplants (as opposed to conservative management), and overall disease state knowledge.6
CDSSs have also been implemented in community pharmacies. A literature review comprised of 6 different studies showed a 31% reduction in drug-drug interactions (DDIs) and a reduced frequency of inappropriate medications for pregnant or elderly patients. The study did, however, find that clinical inertia and alert fatigue were barriers to use of CDSS in the community setting.7
AI is also seen in the ambulatory care setting. A systematic review by Sennesael et al. analyzed 16 different studies involving CDSS and patients requiring anticoagulation.8 Some of the positive features of the CDSS included integration with EMRs, no additional data entry required on the clinician’s side, and recommendations provided (as opposed to only assessments). The study also showed negative features of CDSS such as lack of formalized incentives to use the technology, lack of communication of decision support results to patients, and lack of periodic feedback to clinicians, all of which may be barriers to implementation of CDSS in practice.8
AI can be beneficial to pharmacists in the acute care setting as highlighted in research by Calloway et al., which describes an acute care facility utilizing CDSS to guide pharmacy interventions in antimicrobial stewardship. These interventions involved de-escalation of antibiotic regimens, identification of inappropriate antibiotics based on cultures or patient lab data, dose optimizations, and intravenous to oral conversions. The authors also shared that the CDSS provided alerts to the diabetes educator if abnormal blood glucose values were found in the health records. Utilization of the CDSS therefore increased pharmacists’ clinical interventions by over 100% each month, saving approximately $1.5 million annually.9
Novel medications are currently estimated to cost $3 billion and take up to 10 to 15 years to develop.10 Coupled with the use of in silico models, AI and related techniques (e.g., ML or deep learning [DL]) are able to make predictions across a range of physicochemical properties, pharmacokinetic characteristics, selectivity, and more, which then allows research and development teams to hone in on potentially higher yield preclinical candidate compounds. Deep neural networks (DNN) have also been used in pattern recognition, which can be useful in pharmacology and drug discovery. These DNNs can be used to create novel features beyond the compounds’ traditional molecular structure. However, as DNN algorithms become more commonplace, significant resources will be required to optimize this technology in health care settings.10
AI can be used in drug design to verify protein targets using 3-dimensional structures. Although there is an interest in using AI for this, it remains expensive and time intensive. However, using DL methods in combination with other neural networks has led to advances in target prediction.
Prediction of protein-protein interactions is also possible using AI. This technology may even reduce the rate of adverse events in small molecule drugs due to the increase in selectivity of the drug’s target profile. Furthermore, the neural networks can be used earlier in the drug development process to identify poor absorption, distribution, metabolism, and excretion (ADME) outcomes of the molecule-disease state combination being studied.11
AI is also used for drug repurposing and “hit discovery” by creating disease state models which are validated against either public (e.g., ChEMBL and PubChem) or proprietary (e.g., GOSTAR) chemical libraries.12 Drug repurposing using ML and DL algorithms can be built with an optimized and unbiased data set to authenticate the model performance. These models are then used in conjunction with the previously mentioned chemical libraries with human guidance to further identify the best possible drug candidate in accordance with the disease state.13 These drug models have been used to decrease the time between initial screening and the end of preclinical testing. For example, Exscientia and Sumitomo Dainippon Pharma co-developed DSP-1181, a drug targeting obsessive compulsive disorder. The average time for initial screening to the end of preclinical testing took roughly 12 months, whereas the industry average is 4 to 6 years.14
Pharmacists require the necessary tools and training to understand the fundamentals of AI, whether in the clinical or industry setting. Therefore, more emphasis should be placed on training using avenues such as additions to pharmacy school curriculum, pharmacist continuing education (CE), and digital health research.
AI in Pharmacy Education
As digital health continues to integrate into clinical and industry workflows, health care education must evolve to meet the changing landscape. In order to more seamlessly integrate AI into existing practice settings, digital health education for pharmacists needs to be more accessible. As Aungst et al. discussed in a 2020 piece in the Journal of Medical Education and Curricular Development, there are many proposed models to consider. One potential strategy involves threading digital health topics throughout the pharmacy school curriculum and reinforcing digital health training through elective courses and educational tracks, minors, or certificate programs. Post-graduate opportunities may include master’s degrees, residencies or fellowships, and continuing education credits.15
An exhaustive syllabus search across United States pharmacy schools revealed that few training modalities currently exist, particularly ones involving AI. Several digital health fellowships and permanent roles at well-known universities such as the University of Southern California (USC) and University of California, San Diego (UCSD) have been developed, although a significant gap remains in terms of formal educational opportunities.15
The International Pharmaceutical Federation (FIP) released a comprehensive report on a Global Digital Health Education Framework in May 2021. This report is the first of its kind to showcase global initiatives that integrate digital health into pharmacy education and into the pharmacy workforce and utilized a survey to understand what opportunities already existed.16
The key findings from the survey largely mirrored the US syllabus search, highlighting the limited number of pharmacy schools that offer digital health education.16 It also found a misconception among survey participants, who understood “online education” and “digital health education” to be interchangeable terms. Challenges to offering digital health education were limited resources and availability of experts to guide them in implementation. Despite these findings, there appeared to be a promising trend in that approximately half of participants agreed that their school would be able to incorporate new digital health tools into the curricula as they emerged, and that their students would be prepared to deliver digital health services.
The findings of this report highlighted the knowledge and skill gaps in the pharmaceutical workforce. Implementation of digital health tools in clinical care was one of the least likely concepts covered in pharmacy education, despite participants’ highlighting the clear need to learn how to apply this technology for clinical problem solving. Participants also spoke to the use of digital health technologies such as AI and ML as offering excellent opportunities to improve health care services, access, and outcomes. Participants were most excited about topics such as AI, chatbots, and blockchain technologies, despite having a lower baseline knowledge relative to the more commonly discussed topics of electronic health records and e-prescribing. Existing digital health education seems to include more operational and administrative competencies rather than AI and ML; the report revealed that data interpretation and applications of AI in clinical pharmacy and industry settings seem to be a major gap in digital health training and education.16
There are a few institutions that have successfully implemented digital health training opportunities. Utrecht University in the Netherlands invited students with specific interests in digital health to apply to a pilot elective course. Students had the opportunity to gain diverse exposure to AI, chatbots, and blockchain among other topics through guest speakers, interactive workshops, and applied assignments. Another case study involves a postgraduate course on digital health in Madrid, Spain, which gave students the opportunity to learn about various digital tools, including AI, to help comprehend the power of information technology to deliver more patient-centered care. La Trobe University in Australia was cited for a work-in-progress model on their vertically aligned digital health curriculum that they have integrated into their current pharmacy curriculum. This superimposed model has allowed students starting in their first year to build a foundational framework. In later years, they continue to expand and apply it via existing therapeutics cases and clinical exam role-plays, demonstrating how digital tools can be useful in their practice. They develop deeper understanding through exploring more complex topics such as AI, big data, and robotics. These examples encourage students to stay curious about applying topics such as AI and be at the forefront of their future practice settings.16
Traditional pharmacy school curriculum discusses the clinical fundamentals such as taking a detailed medication history, reconciling medication lists from multiple conflicting sources, and providing thorough patient education. Considering the digital transformation of the health care system and the evolving role of pharmacists, assessing digital health literacy and collecting information on digital technologies used in addition to medication therapies will be a new focus.17 Timothy Aungst writes that “digital health” will just become “health” because of the digital transformation era.18
AI integration in pharmacy practice is inevitable. It is essential that pharmacists understand AI concepts to adapt to future needs. AI-generated clinical decision support can augment the role of the pharmacist as a patient care provider. By interpreting and applying the data provided by the algorithms, pharmacists are uniquely equipped to guide development of more patient-centered care. Educational standards, however, must keep pace with the changes in health care delivery. Pharmacists likely have little to no formal training in AI applications in pharmacy practice given that concepts related to clinical decision making and evidence-based medicine using AI were least likely to be included in pharmacy curriculum, according to survey participants.
During the COVID-19 pandemic, students have been adapting to virtual patient care practices and precepting models to continue their experiential education. There is a need for pharmacists to adopt more digital services beyond telehealth visits in the post-COVID era and implement a digital health competency framework among educators.
Health care has traditionally been a late adopter of the latest digital innovations. Based on the findings from the FIP report, it is essential that digital health education be integrated into pharmacy curricula, especially in AI, ML, and digital medicine. Emphasis should be placed not only on digitally enabling the future generations of pharmacists through the curricular additions, but also on providing support to currently practicing pharmacists through CE and professional development training in emerging digital technologies.
1. Vatanka P. Preparing the Next Generation of Pharmacists for their Role in Digitally Enabled Care. Digital Health presented by CPhA. Published November 8, 2021. https://digitalhealthcpha.com/2021/11/preparing-the-next-generation-of-pharmacists-for-their-role-in-digitally-enabled-care/
2. Amelung K. Pharmacists’ Responsibility in Assessing the Value of Digital Tools for their Patients. Digital Health presented by CPhA. Published September 7, 2021. https://digitalhealthcpha.com/2021/09/pharmacists-responsibility-in-assessing-the-value-of-digital-tools-for-their-patients/
3. Amarasingham R, Plantinga L, Diener-West M, Gaskin DJ, Powe NR. Clinical information technologies and inpatient outcomes: a multiple hospital study. Arch Intern Med. 2009;169(2):108-114. doi:10.1001/archinternmed.2008.520
4. Prescribing Physician Analytics. Arine. https://www.arine.io/prescriber-analytics?hsLang=en
5. Kaiser P, Pipitone O, Franklin A, et al. A Virtual Multidisciplinary Care Program for Management of Advanced Chronic Kidney Disease: Matched Cohort Study. J Med Internet Res. 2020;22(2):e17194. doi:10.2196/17194
6. Dubin R, Rubinsky A. A Digital Modality Decision Program for Patients With Advanced Chronic Kidney Disease. JMIR Form Res. 2019;3(1):e12528. doi:10.2196/12528
7. Curtain C, Peterson GM. Review of computerized clinical decision support in community pharmacy. J Clin Pharm Ther. 2014;39(4):343-348. doi:10.1111/jcpt.12168
8. Sennesael AL, Krug B, Sneyers B, Spinewine A. Do computerized clinical decision support systems improve the prescribing of oral anticoagulants? A systematic review. Thrombosis Research. 2020;187:79-87. doi:10.1016/j.thromres.2019.12.023
9. Calloway S, Akilo HA, Bierman K. Impact of a Clinical Decision Support System on Pharmacy Clinical Interventions, Documentation Efforts, and Costs. Hosp Pharm. 2013;48(9):744-752. doi:10.1310/hpj4809-744
10. Ekins S, Puhl AC, Zorn KM, et al. Exploiting machine learning for end-to-end drug discovery and development. Nat Mater. 2019;18(5):435-441. doi:10.1038/s41563-019-0338-z
11. Zhong F, Xing J, Li X, et al. Artificial intelligence in drug design. Sci China Life Sci. 2018;61(10):1191-1204. doi:10.1007/s11427-018-9342-2
12. Chaturvedula A, Calad‐Thomson S, Liu C, Sale M, Gattu N, Goyal N. Artificial Intelligence and Pharmacometrics: Time to Embrace, Capitalize, and Advance? CPT Pharmacometrics Syst Pharmacol. 2019;8(7):440-443. doi:10.1002/psp4.12418
13. Koromina M, Pandi MT, Patrinos GP. Rethinking Drug Repositioning and Development with Artificial Intelligence, Machine Learning, and Omics. OMICS: A Journal of Integrative Biology. 2019;23(11):539-548. doi:10.1089/omi.2019.0151
14. Burki T. A new paradigm for drug development. The Lancet Digital Health. 2020;2(5):e226-e227. doi:10.1016/S2589-7500(20)30088-1
15. Aungst TD, Patel R. Integrating Digital Health into the Curriculum-Considerations on the Current Landscape and Future Developments. J Med Educ Curric Dev. 2020;7:2382120519901275. doi:10.1177/2382120519901275
16. FIP Digital Health In Pharmacy Education Report Group, Aukje Mantel-Teeuwisse D, Others. FIP Digital Health in Pharmacy Education. Published online 2021.
17. Clark M, Clark T, Bhatti A, Aungst T. The Rise of Digital Health and Potential Implications for Pharmacy Practice. J Contemp Pharm Pract. 2017;64(1):32-40. doi:10.37901/jcphp16-00012
18. Aungst T. 5 Things Pharmacists Should Know About Digital Health. GoodRx Health. Published July 6, 2021. https://www.goodrx.com/hcp/pharmacists/5-things-pharmacists-should-know-about-digital-health