Recent advances in artificial intelligence (AI) have resulted in dramatic changes in the field of medicine. Machine learning (ML) is a type of AI that enables computers to automatically learn from data with minimal human intervention. ML can be divided into three paradigms: supervised, unsupervised, and reinforcement learning. Among them, supervised learning, an ML algorithm that learns the input–output relationship, is most commonly used in medial AI construction.

Despite the recent applications of AI in the medical field, pharmacy faculty education in Japan has not offered AI programs, resulting in gaps between pharmacists’ knowledge about AI and state-of-the-art medical AI technologies.

A new pharmacy education curriculum was launched in Japan in 2024; it emphasizes the importance of acquiring knowledge and skills to understand and use AI. With the growing use of AI in the medical field, pharmacy students and pharmacists in practice must acquire AI-related knowledge and skills.

In this review, we outline the methodologies for applying supervised learning. We also provide code examples for supervised learning to help beginners implement AI.

Nutritional indices have been reported as predictors of therapeutic efficacy in Atezolizumab + Bevacizumab combination therapy (Atezo + Bev therapy) for unresectable hepatocellular carcinoma; however, the number of such reports is limited. Therefore, we investigated whether each nutritional index could be a predictor of therapeutic efficacy and compared their usefulness. The relationship between therapeutic efficacy and changes in nutritional indices was also examined. A total of 62 patients with unresectable hepatocellular carcinoma who initially received Atezo + Bev therapy were divided into a stable disease (SD) or greater group (40 patients) and a progressive disease (PD) group (22 patients); nutritional indices at the start of the treatment were compared. For nutritional indices that showed significant differences (neutrophil-to-lymphocyte ratio [NLR], platelet-to-lymphocyte ratio [PLR], prognostic nutritional index [PNI], and cholesterol-modified prognostic nutritional index [CPNI]), cutoff values (NLR: 2.3, PLR: 118.2, PNI: 41.0, and CPNI: 86.9), predictive ability, and area under the curve (AUC) were calculated. As a result, PLR and CPNI demonstrated the highest accuracy of 75.8％, with an AUC of 0.71. Patients were classified into two groups using the cutoff values of PLR and CPNI, and the progression-free survival (PFS) and overall survival (OS) were compared. The PFS and OS were significantly longer in the group with a good nutritional status. Changes in PLR and CPNI between the start of the treatment and the time of PD assessment were compared for 22 patients in the SD or better group who developed PD during the observation period; however, no significant changes were observed.

In pediatric pharmacotherapy, the bitterness of the medication often hinders effective treatment. Sulfamethoxazole-trimethoprim (ST) combination drugs are commonly used to prevent infections in immunocompromised patients but are very bitter, making them difficult for children to consume. To address this issue, we developed a jelly formulation from crushed ST tablets that can be prepared at hospital pharmacies. We evaluated the bitterness masking effect 24 hours after preparation using a sensory test with a Visual Analogue Scale (VAS) in healthy adults. The test compared “water,” “ST tablet suspension,” “150-fold diluted suspension,” and “jelly formulation containing ST tablets.” In the bitterness test, the bitterness VAS score of the “jelly formulation containing ST tablet” was significantly suppressed compared to the “ST tablet suspension,” indicating that jellification reduced the bitterness of the ST tablet. In addition, we prepared the jelly and measured its hardness over time using a creep meter. The jelly formulation containing the ST tablet could be packaged in portion cups within 30 minutes after preparation (approximately 2000 Pa), and after 1 day, the jelly strength was sufficient to be easily scooped up with a spoon (approximately 13000 Pa). The jelly formulation we developed is easy to handle at hospitals and pharmacies and reduces bitterness. By switching from tablets to jelly formulations, bitterness was reduced, making it more child-friendly and easier to consume, thereby expanding the possibility of options for medication for pediatric patients.

Acetaminophen (APAP) is a major drug that causes acute poisoning, including severe liver damage. In cases of overdose, the serum APAP concentration is used as an indicator for administering the antidote N-acetylcysteine (NAC). However, measuring APAP levels in clinical settings is often challenging; a threshold of 7,500 mg or 150 mg/kg is commonly used for NAC administration. This guideline was developed for Westerners, and its applicability to Japanese individuals is unclear. This study aimed to explore factors other than serum APAP concentration that could quickly lead to NAC administration. We analyzed patients (n = 25) who visited the emergency department of Ogaki Municipal Hospital and had their APAP levels measured between April 2014 and March 2024. Patients were divided into NAC-recommended and non-recommended groups based on the NAC administration line in the Rumack–Matthew nomogram, and APAP intake amounts were compared. The cutoff value was calculated using the receiver operating characteristic (ROC) curve. The median APAP intake (interquartile range) was 9,500 (7,380 – 14,400) mg in the NAC-recommended group and 4,000 (3,925 – 6,250) mg in the non-recommended group (P = 0.045). The calculated cutoff value using the intake curve was 5,200 mg (120 mg/kg). The area under the ROC curve for APAP intake was 0.792, indicating moderate accuracy in event detection (95％ confidence interval: 0.583 – 1), with sensitivity and specificity for detecting NAC administration at 91.7％ and 70.0％, respectively. This study suggests that Japanese patients who ingest more than 5,200 mg or 120 mg/kg of APAP may qualify for NAC administration.

A risk management plan (RMP) is essential for evaluating and minimizing the risks associated with pharmaceuticals throughout their lifecycle. However, previous research has indicated that its adoption in community pharmacies is limited. In the fiscal year 2024 revision of the medical service fee, “Specific Drug Management and Guidance Fee 3 (Category I)” was newly established. This fee is reimbursable when a patient handbook related to the RMP is provided to a patient who is prescribed a medication covered by the RMP for the first time. This study investigated RMP utilization in community pharmacies following the establishment of Specific Drug Management and Guidance Fee 3 (Category I). A questionnaire survey was conducted among pharmacists-in-charge at 626 community pharmacies in Iwate Prefecture from December 2024 to January 2025. A total of 138 pharmacists (22％) responded, and their responses were analyzed. More than half of the respondents reported increased RMP understanding and utilization following the establishment of Specific Drug Management and Guidance Fee 3 (Category I). The most common RMP application was the utilization of patient handbooks for medication counseling, identifying risks when initiating the handling of new drugs, and evaluating adverse effects. To further enhance RMP utilization, respondents emphasized the importance of providing concrete examples, integrating RMP information into electronic medication records, and conducting training sessions. In conclusion, the establishment of Specific Drug Management and Guidance Fee 3 (Category I) promoted RMP utilization in community pharmacies.

The role of pharmacists in providing home medical care has gained increasing importance in recent years. At Asanogawa General Hospital, pharmacists provide home medical care services when requested by physicians; however, the evaluation of these services by other healthcare professionals remains unclear. Therefore, this questionnaire survey aimed to assess other healthcare professionals’ perceptions and evaluations of hospital pharmacist interventions, and their expectations of these services, to improve future home medical care services. The participants included medical professionals and welfare/caregivers, such as care managers. The results were compared between those with and without prior experience in requesting home medical care services from hospital pharmacists; free-text responses were analyzed using text mining. A total of 95.1％ of respondents reported satisfaction with the home medical care services provided by hospital pharmacists. Among the reasons for requesting home medical care services, the most common expectation was improvement in medication management. The statistical analysis revealed that awareness of home medical care services provided by pharmacists was significantly higher among those who had previously requested such services. The free text responses also indicated a strong desire for enhanced collaboration and information sharing with other healthcare professionals. The findings of this survey indicate that the services provided by hospital pharmacists are highly valuable. Our findings suggest the need for broader dissemination of information regarding pharmacists’ involvement in home medical care, along with improved coordination with community pharmacists.