Skip to content

USING ARTIFICIAL INTELLIGENCE TO TRAIN FUTURE PHARMACISTS : EVALUATION OF A CHATBOT DEVELOPED FOR THE ONCOLOGY PHARMACY UNIT

European Statement

Education and Research

Author(s)

P. BROUARD
C. JUTARD
C. COUSIN
E. COGET

Why was it done?

The Oncology Pharmacy Unit within the hospital pharmacy is experiencing a steady increase in activity. Medical teams are increasingly calling on pharmacy interns for technical and clinical questions.

What was done?

To enhance the training of pharmacy interns and improve responsiveness to medical inquiries, an artificial intelligence (AI) driven pharmaceutical chatbot pre-configured using a dedicated prompt. It was developed to deliver accurate, context-specific answers on clinical pharmacy, pharmaceutical technology, and organizational aspects of the oncology unit. This study aims to assess the relevance and quality of its responses to evaluate its potential as a support tool in intern training.

How was it done?

A total of 123 anonymized documents, including procedures, educational materials, and internal resources from the oncology unit of the hospital pharmacy, were integrated into a dedicated chatbot developed with ChatGPT-5. Forty questions reflecting eight key thematic areas of residents’ activities—clinical and technological validation, outpatient dispensing, controlled atmosphere areas, medical staff inquiries, software use, routine practice, and on-call situations—were submitted to the AI. Four experts pharmacists independently assessed each response using a standardized evaluation form with four levels of agreement (“Agree,” “Somewhat agree,” “Somewhat disagree,” “Disagree”).

What has been achieved?

Among the 160 responses received, a strong inter-rater consistency was observed, with 90% of evaluations showing concordance among at least three assessors and full unanimity in 20% of cases. Overall, 79% of chatbot responses were rated favorably (41% “Agree,” 38% “Somewhat agree”). The highest satisfaction rates were found in clinical pharmacy (95%), controlled atmosphere areas (95%), and software (90%). In contrast, lower agreement occurred in pharmaceutical technology (65% favorable, 30% disagreement) and particularly in on-call situations, where 55% of responses were considered unsatisfactory.

What next?

These findings confirm the educational potential of this tool, with most answers deemed relevant. Nonetheless, discrepancies underline its limitations regarding specific technical data, including occasional hallucinatory, incorrect, or incomplete responses that lack adequate reference to institutional procedures. Given its potential, we are working to improve this chatbot by refining the prompt, updating the database, and testing other non-hallucinatory AI models.

STRUCTURED INNOVATION GOVERNANCE: AN INTERPROFESSIONAL COMMITTEE IN HOSPITAL PHARMACY

European Statement

Introductory Statements and Governance

Author(s)

Vanusa Barbosa Pinto, Cleuber Esteves Chaves, Andréa Cássia Pereira Sforsin, , Priscila Faria França, Mayara Araújo Dias, Erik Magnus Lindh, Caroline Sandoli de Almeida Souza, Maria Cleusa Martins, Maristela Barros De Sousa, Rafael Alves de Souza,

Why was it done?

We implemented a structured, interprofessional Innovation Committee within the hospital pharmacy to systematically manage the entire innovation pipeline. The team, comprising pharmacists, nutritionists, physicians, and Information Technology (IT) specialists, established a formal process to guide high-potential projects from initial ideation to final submission for competitive funding. This governance model was successfully applied in 2025 at the pharmacy of a public teaching hospital.

What was done?

Innovation often lacks strategic coordination in hospital pharmacy, limiting the translation of valuable ideas into robust projects. Our objective was to overcome this unstructured environment by creating a governance framework. The committee began its work by specifically focusing on identifying deep clinical “pain points,” such as fragmentation in antimicrobial management, difficulty in customizing medication dosages, and low adherence to training programs for Generation Z staff.

How was it done?

Projects were prioritized based on a methodology that weighed clinical impact, economic feasibility, and technical executability. The team utilized agile management tools, including the value-versus-effort matrix and the problem-solution canvas, complemented by sprint rituals to ensure progress and alignment. The committee successfully generated and developed three large, scalable proposals, validating the model’s capacity to identify and mature high-impact ideas. These proposals were submitted to a competitive institutional innovation grant (In.Cube-InovaHC).

What has been achieved?

The structured process resulted in a robust innovation pipeline with three high-potential proposals: PrintPharma (3D-printed personalized medications), FarmáciaLab (a gamified platform for team training), and Sentinela-ATB (an antimicrobial stewardship hub). The PrintPharma project, which aimed to develop an in-hospital 3D printing solution for personalized medicines, was ranked 8th among 134 highly competitive proposals in a major institutional innovation grant (In.Cube-InovaHC). This ranking validated the quality and maturity of the committee’s output.

What next?

This structured, pharmacy-managed innovation pipeline is a feasible and high-impact strategy that significantly strengthens the institution’s capacity to drive change. It should be considered a best practice example because the governance model and its agile tools are fully replicable and adaptable by any other hospital pharmacy, establishing the pharmacist as a protagonist in healthcare innovation.

ELECTRONIC PRESCRIPTION INTEGRATION: ACHIEVING 98% COVERAGE AND REDUCING TRANSCRIPTION ERRORS

European Statement

Patient Safety and Quality Assurance

Author(s)

VB Pinto, CSA Souza, NL Mizutani, RP Santos, MA Dias, ACP Sforsin

Why was it done?

The unification of two outpatient prescription platforms into a single, institutional electronic system was carried out. The initiative was led by a multidisciplinary team (pharmacists, physicians, and systems analysts) from a public, tertiary care hospital.

What was done?

Despite the high rate of electronic prescribing (92%), the coexistence of the two systems generated data fragmentation, rework, and exposed patients to unnecessary risks of transcription errors. The central objective was, therefore, to unify the process to universalize the use of electronic prescribing and, consequently, reduce transcription errors, strengthening patient safety and pharmaceutical care.

How was it done?

The intervention took place between July and September 2025. Actions included the technical integration of the systems into a single platform, the development of new standardized prescribing protocols, and training for prescribers. The impact was evaluated by monitoring the electronic prescribing coverage percentages and transcription errors rates at the outpatient level before and after implementation.

What has been achieved?

The integration demonstrated fast and significant results. Electronic prescription coverage increased from 92% to 98% in just one month after unification. In parallel, there was a progressive and sharp reduction in transcription errors: from 8.5% (pre-implementation) in July to 2.7% in September (after integration), representing a 68% decrease in the manual correction rate.

What next?

This initiative demonstrates the importance of technological unification for the advancement of Hospital Pharmacy. The model is highly replicable for other institutions dealing with fragmented electronic prescribing systems. The integration of systems has a direct and measurable impact on patient safety, establishing universal electronic prescribing as an efficient care standard and proof against transcription errors.

CYBER ATTACK! IS YOUR PHARMACY TEAM READY?

European Statement

Patient Safety and Quality Assurance

Author(s)

Sandra Oliver, Chief Clinical Information Officer, Pharmacy and Medicines.
Chris Green, Director of Pharmacy and Medicines Optimisation

Why was it done?

We describe how our hospital experienced a cyber-attack which resulted in the electronic patient record and other systems being taken offline, including our electronic prescribing system.

What was done?

We describe our response to the cyber-attack and the considerations necessary to replace an entire hospital’s electronic prescribing system with paper. We also describe the co-dependencies with other previously electronic departments and systems, and communications with partner organisations.

How was it done?

The conversion from electronic to paper prescribing was led by the pharmacy team. Initially, this was via a printed downtime solution from the electronic prescribing system, which then required transcription to a paper prescribing chart. This generated a number of logistic, practical and safety issues which were managed by the pharmacy team. We then describe how the hospital was converted back to electronic prescribing over the course of one day, and how the pharmacy team was pivotal to that.

What has been achieved?

We were able to successfully take our hospital off electronic prescribing and initiate a paper-based downtime solution that ran over several days, and then return the hospital to electronic prescribing. Recovery efforts involved pairing prescribers with pharmacists across wards, establishing a pharmacy command centre, coordinating medication administration on ePMA, and communicating service updates across the Trust. Third-party services were reinstated, and recovery was prioritised by acuity, with the least acute wards addressed first. During the downtime period, we actively encouraged the pharmacy team to log patient safety concerns, or ideas to improve the downtime process which were reviewed and acted on in real time and as part of the post-recovery review.

What next?

We have learned several valuable lessons from the downtime experience which we have taken forward as part of our revised business continuity plan. This includes a review of specialist prescribing situations for example insulin, how we contact and work with partner organisations, the robustness of our downtime printer setup, and training material for clinical staff who at the time of the cyber-attack, had never used a paper based prescribing system. This case study serves as a valuable resource for healthcare organisations seeking to strengthen their cybersecurity and business continuity strategies.

OPTIMIZATION OF PATIENT RECRUITMENT IN CLINICAL TRIALS THROUGH ARTIFICIAL INTELLIGENCE INTEGRATION (TRIALGPT PROJECT)

European Statement

Education and Research

Author(s)

Laura Maldonado Yagüe, Claudia Ramos Álvarez, Ana Herranz Alonso, Fernando Bustelo Paz, Eva González-Haba Peña, María Sanjurjo Saez

Why was it done?

Patient recruitment is still nowadays one of the barriers that the clinical investigation encounters: almost 80% of the clinical trials experiment delays and around 30% close due to the difficulties to identify candidates. Currently, the recruitment process in many of the hospital sites is based on the manual review of electronic medical records (EMR), which results in higher workload and higher errors and omissions. This tool aims to reduce manual screening time by 50% and increase recruitment by 20% always ensuring regulatory compliance (ICH-GCP, RGPD and national biomedical investigation laws).

What was done?

The Pharmacy Department led the initiative to evaluate TrialGPT, an Artificial Intelligence (AI) system designed to optimize clinical trials patient recruitment, in the hospital setting. Natural Language Processing (NLP) and Large Language Model (LLM) are advanced techniques used by TrialGPT which enables the automatic detection of potential eligible patients through their Electronic Medical Records (EMR) matching their profile to the inclusion and exclusion criteria for each clinical trial.

How was it done?

The project was coordinated by a Clinical Trials Unit of a tertiary hospital with multidisciplinary collaboration between pharmacist, investigators and IT specialists. Anonymized data of 50 active clinical trials from oncology, neurology and rare diseases areas were used. Technical challenges such as data heterogeneity, algorithmic bias and staff acceptance were encountered, in order to address these, an iterative training model, multidisciplinary workshops and ethical evaluation were used.

What has been achieved?

Preliminary simulations indicate that TrialGPT is able to reduce half the necessary time for patients screening and improve recruitment efficiency without compromising data security or clinical precision. The model achieved high sensitivity and specificity identifying eligible patients, demonstrating a high potential to optimize hospital investigation flowcharts.

What next?

A validation phase will evaluate the real-world performance and scalability in multiple sites. This initiative exemplifies an innovative digitalization and automatization of a process which could be transferred as a model for European hospitals in order to improve patient access to clinical trials, thus, advanced therapies.

BUILDING A ZERO-ERROR MEDICATION WORKFLOW THROUGH SMART DISPENSING SYSTEMS AND BIG DATA GOVERNANCE

European Statement

Patient Safety and Quality Assurance

Author(s)

Hui-Yu Chen, Kai-Cheng Chang

Why was it done?

Medication safety remains a cornerstone of healthcare quality, yet adverse drug events (ADEs) continue to cause preventable harm in hospitals. Traditional manual dispensing workflows, dependent on human memory and paper-based checks, are prone to errors, particularly in large medical centers with high prescription volumes.

What was done?

We launched the “Smart Dispensing and Data Governance Project,” aiming to transform the pharmacy workflow through digitalization and data-driven quality management.

How was it done?

A two-pronged strategy was adopted: (1) deployment of smart dispensing hardware and (2) establishment of a big data governance platform. The hardware featured personalized login for accountability, closed-loop barcode verification of both medications and prescription bags, LED guidance and voice feedback for real-time alerts, and final barcode validation before dispensing. Advanced automation such as real-time stock sensing, weight-based verification, and image-assisted accuracy checks further minimized human errors. A SAS Visual Analytics–based Business Intelligence dashboard visualized error trends and enabled continuous PDCA (Plan-Do-Check-Act) improvement cycles through near real-time feedback.

What has been achieved?

Implementation led to substantial quality improvements: the dispensing error rate decreased by 78.3% (0.023‰ to 0.0050‰, P < 0.05); data analysis time for error monitoring shortened from 4 hours to 10 minutes (-98.3%, P < 0.05); and pharmacist training time reduced from 10 days to 3 days (−60.0%). All indicators showed statistically significant enhancement in accuracy and efficiency. Integrating smart dispensing systems with big data governance effectively advanced medication safety and operational efficiency. This model established a scalable, data-driven, and high-reliability pharmacy workflow, transforming quality management from reactive correction to proactive prevention and serving as a replicable benchmark for digital hospital transformation.

What next?

We plan to apply AI algorithms to dynamically optimize drug storage locations based on usage and safety risk, and to digitalize all storage displays through an integrated electronic shelf–label system. These enhancements will further strengthen accuracy, reduce human-factor variability, and advance a highly reliable smart dispensing workflow.

SOFTWARE FAILURE IN A RADIOPHARMACY: IMPLEMENTATION OF A BUSINESS CONTINUITY PLAN AND DEVELOPMENT OF SUPPORT TOOLS

European Statement

Patient Safety and Quality Assurance

Author(s)

A. BRUYÈRE1, A. DESCHAVANNES1, C. RIOUFOL2, M. PIQUEMAL1.
1GROUPEMENT HOSPITALIER SUD – HOSPICES CIVILS DE LYON, RADIOPHARMACY, LYON, FRANCE.
2GROUPEMENT HOSPITALIER SUD – HOSPICES CIVILS DE LYON, PHARMACY, LYON, FRANCE.

Why was it done?

Following an IT failure in our radiopharmacy and nuclear medicine departments, the professional software system became unavailable for over two hours. This incident led to the establishment of a Business Continuity Plan (BCP) to ensure preparedness for similar future events. The BCP aims to guarantee the resumption and continuity of operations in the event of disruptive situations.

What was done?

This study aims to ensure operational continuity and safety of radiopharmaceutical workflows during IT failures through the development of a BCP and support tools.

How was it done?

A multidisciplinary staff meeting, using a Focus Group approach, was organized to reach a consensus on service organization during degraded operating conditions. The workflows of both radiopharmaceuticals (RPs) and patients were analyzed. A feedback analysis of the initial incident was also conducted to adjust the proposed tools to daily operational constraints.

What has been achieved?

A specific procedure was developed, detailing the overall organization of the department during degraded conditions and defining the tasks of each professional group: patient reception, paper-based prescription, scheduling of injections and imaging sessions, RP preparation, quality control, dispensing, and traceability. Paper versions of all necessary documents were compiled into a “BCP binder,” including pre-filled labels, RP quality control batch records, prescriptions, and dispensing registers. A preparation support chart for RP activities was also included. Additionally, digital tools were designed to support dispensing, such as an Excel spreadsheet allowing automated decay calculations and volume determinations for each dispensation.

What next?

This organizational work, along with the development of both paper-based and digital tools, has strengthened the safety and continuity of the RP circuit during IT outages. However, real-life testing of the BCP is required to confirm its feasibility. Given their reliance on digital systems, professionals may face adaptation challenges in crisis situations, increasing the risk of error. Regular training sessions and simulation exercises are therefore essential to ensure BCP effectiveness. In parallel, a broader BCP is being developed to address other identified risks, including cybersecurity threats, network failures, and equipment or facility malfunctions (e.g., automated system breakdowns or air supply interruptions).

INTEGRATING ARTIFICIAL INTELLIGENCE: STREAMLINING MEDICATION HISTORY DOCUMENTATION WITH CHATGPT AT ODENSE UNIVERSITY HOSPITAL

Pdf

PDF Icon

European Statement

Clinical Pharmacy Services

Author(s)

Fjóla Høg Nielsen, Gine Cecilie Stobberup

Why was it done?

The initiative was driven by the need to support a growing number of patients who require medication history. To ensure that pharmacy technicians could complete as many medication histories as possible, have enough time to thoroughly investigate potential issues, and maintain the desired quality of medication history, we initiated the use of ChatGPT to handle the journal note-writing aspect. This initiative was launched in January 2024 and implemented in August 2024.

What was done?

At Odense University Hospital, pharmacists and pharmacy technicians conduct daily reviews of hospitalized patients’ regular medication based on data from the Danish Medicines Agency’s system “The Shared Medication Record”, prescription deliveries, and patient statements. The patient’s usual medications are reviewed to determine what they are taking at home and to identify potential issues, such as compliance problems. A note is written in the medical journal for the attending physician, providing an overview of the patient’s regular medications and any concerns. To improve efficiency and consistency in this process, we implemented the use of ChatGPT to write these notes after the pharmacy technician has completed the medication history, ensuring standardized documentation, increased safety, and for saving time.

How was it done?

ChatGPT was programmed to document the medication history following the standard note format previously used. One of the key challenges was ensuring that ChatGPT could meet the specific documentation needs and minimizing errors in the generated notes. After the initial programming, pharmacy technicians were asked to use ChatGPT and keep track of how many medication histories were written with its assistance, as well as to identify any recurring errors. Based on their feedback, ChatGPT was adjusted to reduce the occurrence of similar errors in future notes.

What has been achieved?

Pharmacy technicians have reported that the time required to complete a medication history has decreased, particularly for patients with long medication lists. The system also ensures that the notes are always written in a consistent manner, reducing the likelihood of missing important information.

What next?

Moving forward, we will continue refining the system to further eliminate errors and improve accuracy. This initiative showcases the successful integration of advanced technology into healthcare, with potential applications across other healthcare settings.

IMPLEMENTATION OF AN ATYPICAL MEDICATION ROOM TO OPTIMIZE MEDICATION MANAGEMENT AND REDUCE WASTE AT HERLEV GENTOFTE HOSPITAL, DENMARK

Pdf

PDF Icon

European Statement

Clinical Pharmacy Services

Author(s)

Henrik Kjer, Christina Laustsen, Rasmus Riis, Caroline Rasmussen, Jeanette Bajrami, Christian Rubek, and Steffen Jørgensen

Why was it done?

An atypical medication room (AMR) was established at Herlev Gentofte Hospital, Denmark to centralize the storage and handling of medications not part of the standard assortment (i.e. atypical medication). The project aimed to improve the efficiency of medication management, reduce medication waste, and streamline workflows associated with the use of atypical medicines. To enhance the accuracy and efficiency of inventory control, the ScanPill technology was developed as a tool for digital tracking and updating of medication stock.

What was done?

Atypical medications are often stored across various departments with low turnover, leading to potential waste and time-consuming retrieval processes. Centralizing these medications in an AMR and using ScanPill aimed to reduce waste due to expiry, improve stock management, and simplify medication retrieval for healthcare professionals.

How was it done?

Atypical medications from multiple departments were collected and stored in the AMR. The ScanPill system was developed to facilitate the scanning of QR codes and barcodes on medication packaging, allowing for precise tracking of stock levels and easy updates to the atypical medication list. Staff were trained to use the AMR and ScanPill to ensure smooth transitions in retrieving, returning, and documenting atypical medicines. Regular inventory checks and updates were conducted to maintain an accurate database of available medications.

What has been achieved?

The AMR, supported by ScanPill, led to improved handling and management of atypical medications. The centralized storage reduced the need for duplicate stock across departments and enabled quicker access to necessary medications, reducing retrieval time and potential waste. The ScanPill technology improved inventory accuracy and streamlined the process of checking medication in and out, ensuring up-to-date records. Staff feedback has been positive, noting enhanced workflow efficiency and reduced medication waste.

What next?

Future steps include evaluating the economic impact of the AMR and its effectiveness in reducing medication waste. Efforts will be made to refine the use of ScanPill, enhance staff training, and explore potential applications of the AMR model across other departments. Continuous monitoring will ensure optimal performance and identify further areas for process improvement.

VIDEO CONTENT: FEEDBACK ON THE IMPLEMENTATION AND THE MANAGEMENT OF AN AUTOMATED CYTOTOXIC PRODUCTION UNIT

Pdf

PDF Icon

European Statement

Production and Compounding

Author(s)

M. SIEGWART, A. BENDJAMA, D. KAROUBY, T. MARTIN, L. CITTADINI, MG. MARTINS, P. COLIAT

Why was it done?

Two automated preparation robots were implemented at ICANS in a context of increased activity in oncology, the need to maintain quality assurance in the preparation process and to reduce pharmacy technician’s exposure to cytotoxic agents. Preliminary professional training is crucial to understand this new technology, master the equipment and interfaces, and adapt to the new circuit and procedures.

What was done?

This work was the development of an educational virtual tour of an automated production unit, enriched by feedback, accessible to any professional interested in implementing an automated preparation robot.

How was it done?

The project was developed in collaboration with the Grand Est regional oncology network (NEON). Scripts were written based on a plan, detailing texts and scenarios to create short videos, each addressing a different theme with a voice-over narration. A professional team from NEON shot the film and edited according to the scripts. Location scouting and filming were completed over 3 days, with voice-over recording and editing done afterward.

What has been achieved?

Six scripts were produced. The first introduces the centre, while 4 others detail the management of an automated unit, including the organization and operation of storage areas, decontamination SAS, personal SAS, and the cleanroom (functional parameters, particulate class, airflow schema, dressing and hygiene rules, microbiological controls, cleaning, and the composition of the “breakage kit”). The robots are covered in a dedicated script that discusses the context of automation, their operation, the software used, possible interfaces, preparation procedures, and cleaning. The final script reviews the entire circuit, linking each area and stage of production: pharmaceutical validation, automated production management, material and vial preparation, manufacturing, and pharmaceutical release.

What next?

The virtual tour presents the circuit and the role of each involved personnel, highlighting precautions and subtleties compared to a non-automated circuit according to the most recent french guidelines. Although the practices shown may not be applicable to all centres due to differences in production area layouts, structures, staffing, and equipment, these videos aim to clarify the functioning of an automated unit while adhering to the guidelines. This online training can promote the standardization of practices, helping professionals from other centres install production automation systems. It encourages innovation and supports pharmacists during this critical transition.