AI Artificial Intelligence in Pharmaceutical Galenic Field: Useful Instrument and Risk Consideration

    

    

    Figure 4: Adapted from AI-based root cause analysis of Hans-Ulrich
Kobialka 2022 [7].
    

Materials and Methods

This study employed an observational approach to review scientific literature pertinent to the integration of artificial intelligence (AI) in the pharmaceutical and galenic fields. Relevant articles were identified and analyzed to understand the current landscape of AI applications and their implications for pharmacy practice. Figures 1 to 5 were included to visually represent key concepts and findings discussed in the literature.

Additionally, a practical component was incorporated into the study, wherein a series of queries—ranging from simple to complex— were submitted to a well-known AI chatbot. The chatbot’s responses were documented and evaluated for accuracy and relevance to pharmaceutical practice. This practical experience aimed to provide insights into the capabilities and limitations of AI technology in addressing specific pharmaceutical inquiries.

Following the literature review and practical experience, a comprehensive conclusion was drawn to summarize the findings and implications for researchers and practitioners in the field.

Results

Literature Review Findings

Cognitive Limitations in Medicine: Aliasghar Karimi et al. noted that the human mind faces numerous obstacles in recalling and applying vast amounts of medical information. The proliferation of medical knowledge makes it impractical for clinicians to analyze extensive literature, leading to diagnostic errors primarily attributed to cognitive biases among healthcare workers. Medical errors remain a significant cause of mortality in the United States, often linked to human error [8].

AI in Pharmacy Practice: Sri Harsha Chalasani et al. highlighted AI as a transformative technology capable of enhancing medication management and patient care. By leveraging AI algorithms and machine learning, pharmacists can analyze extensive patient data, improving the identification of drug interactions and informing tailored recommendations [9].

Personalized Medicine: Lalitkumar K. Vora et al. emphasized that AI can facilitate personalized medicine through the analysis of real-world patient data, leading to improved treatment outcomes and adherence [10].

Streamlining Clinical Decisions: Kelsee Tignor et al. discussed how AI, referred to as pharmacointelligence, can assist clinical pharmacists in making evidence-based decisions by analyzing large volumes of patient data [11].

Enhancing Patient Safety: Rayn Oswalt noted that pharmacists are concerned about patient safety and that AI can help detect and prevent medication errors, thereby reducing adverse events and hospital readmissions [12].

Applications in Dosage Form Development: Praveen Halagali et al. reviewed AI’s applications in developing solid dosage forms, optimizing formulation processes, and assessing drug toxicity profiles, which streamline the path from pilot studies to market [13].

Excipient Compatibility: Ashutosh Kumar et al. pointed out that AI can significantly enhance the assessment of excipient compatibility, improving pharmaceutical development [14].

Predicting Drug Toxicity: Mahroza Kanwal Khan et al. highlighted the advantages of using AI to predict drug toxicity, which allows for a deeper understanding of drug interactions with biological systems [15].

Formulation Optimization: Negar Mottaghi et al. discussed how AI algorithms evaluate data to enhance the stability and compatibility of pharmaceutical ingredients, leading to improved formulations [16].

Pediatric Dosing: Andreea-Alexandra Mocrii et al. aimed to assist pediatricians in determining appropriate treatment doses for children based on various parameters [17].

Education on AI in Pharmacy: Muhammad Ahmer Raza et al. advocated for pharmacy education to incorporate AI and data science, emphasizing the need for pharmacists to develop skills that promote AI integration in practice [18].

Systematic Review of AI in Healthcare: Margaret Chustecki reported a systematic review yielding 8796 articles, which narrowed down to 44 studies. The review highlighted AI’s potential in healthcare, such as improving diagnoses and personalized treatment plans, while also addressing concerns about biases and data privacy [19].

Risk Prevention in Clinical Practice: Michela Ferrara et al. emphasized the usefulness of AI in risk prevention and incident reporting in clinical settings [20].

Advancing Toxicology: Nicole Kleinstreuer et al. noted that AI could advance toxicology into a more predictive discipline, safeguarding human and environmental wellbeing [21].

Risks Associated with AI: Mateusz LASKA et al. identified human error as a significant risk in AI applications, emphasizing the need for understanding and managing these technologies [22].

Ethical Considerations: Mitul Harishbhai Tilala et al. discussed the multifaceted ethical considerations surrounding AI in healthcare, including privacy, bias, and accountability [23].

3D Printing in Pharmaceutical Development: Timothy Tracy et al. highlighted the versatility of 3D printing technology in creating customized dosage forms, which accelerates formulation development [24].

Prescription Monitoring: Cinzia Barberini et al. described a system that interconnects prescription-related aspects, enhancing stock monitoring and preparation checks [25].

AI System Failures: Sasanka Sekhar Chanda et al. identified potential failure points in AI systems, including input errors and processing deficiencies, which can lead to inappropriate outputs [26].

Risks of AI in Medicine: Karim Lekadir et al. outlined seven main risks of AI in healthcare, including patient harm, misuse of tools, and gaps in accountability [27].

Chatbot Accuracy in Healthcare: Stefanie Beck et al. evaluated the accuracy of ChatGPT versions 3.5 and 4 in healthcare contexts, revealing discrepancies in adherence to established guidelines [28].

Variability in Chatbot Responses: Meron W. Shiferaw et al. noted inconsistencies in ChatGPT’s responses, highlighting errors that could result in clinical harm [29].

Performance Comparison of ChatGPT Versions: Ronald Chow et al. reported varying performance levels between ChatGPT versions, emphasizing the need for further development to enhance reliability in medical training and decision-making [30].

These findings underscore the transformative potential of AI in pharmacy and healthcare while also highlighting the associated risks and ethical considerations that must be addressed to ensure safe and effective implementation.

Experimental Project

In this section, a series of queries—ranging from simple to complex—were submitted to a widely recognized AI chatbot accessible for free on the web. The responses received are documented below:

1. Is digoxin water-soluble?Response: It is poorly soluble in water and more soluble in alcohol.

2. What is the molecular weight of NaCl?Response: 58.44 g/ mol.

3. Must collyria be sterile?Response: Yes, it is fundamental to avoid eye infections.

4. Is propranolol compatible with microcrystalline cellulose?Response: Yes.

5. Is omeprazole acid-sensitive?Response: Yes, it is inactivated in the acidic environment of the stomach.

6. What is the chemical structure of citric acid monohydrate?Response: (The chatbot did not provide a specific right structure).

7. Is lidocaine hydrochloride considered a poison by the Italian Pharmacopoeia No. 3?Response: Yes.

8. Is acetone flammable?Response: Yes.

9. What is the galenic use of NIPAGIN?Response: Preservative.

10. What is the water solubility of amoxicillin?Response: The water solubility is not extremely high, so specific formulations are needed to improve body absorption.

11. How can the solubility of water-insoluble active principles in galenic oral drops be increased?Response: Various strategies include using cosolvents like ethanol, glycerin, and propylene glycol; employing surfactants; utilizing cyclodextrin complexation; applying solubility enhancers (e.g., PEG); adjusting pH; and formulating suspensions.

12. How can the solubilization of a solute in a solution be increased?Response: Increase temperature, use a suitable solvent, apply agitation, increase the surface area of the solute, employ surfactants, adjust the pH, use co-solvents, apply pressure (for gases), and utilize complexants.

13. How many grams of KOH are needed to prepare a 100 ml solution at 30%?Response: Dissolve 30 grams in a total volume of 100 ml of solution.

14. How should medications be administered to children with difficulty swallowing?Response: Liquid suspensions, chewable tablets, dissolvable forms, powders, orally disintegrating tablets, other liquid forms, or suppositories.

15. What is the time needed to adequately mix APIs and excipients in the galenic field using a semi-automatic powder mixer to prepare capsules?Response: The chatbot provide answer as reported in the Figure 5.

    

    

    Figure 5: Response from an AI Chat bot.