INTRODUCTION:

Ensuring the efficacy, safety, and consistency of drug formulations is paramount¹. This responsibility relies heavily on rigorous quality control tests, which assess the performance of active pharmaceutical ingredients (APIs) such as paracetamol, a widely used analgesic and antipyretic. Calibration curves and dissolution tests are among the most important quality control procedures^2–5.

These tests allow researchers to determine the concentration of a drug in solution and assess the rate at which it is released from a solid dosage form. Accurate results from these tests are vital to guarantee that a drug will deliver the intended therapeutic effect and meet regulatory standards².

While artificial intelligence (AI) has made significant strides across various industries, its role in pharmaceutical quality control remains largely exploratory. AI tools, such as ChatGPT, offer intriguing possibilities for generating simulated data, aiding in data analysis, and predicting experimental outcomes. These tools have the potential to enhance efficiency, reduce costs, and offer educational value by simulating laboratory processes^7–9. However, the complexities and precision required in real-world pharmaceutical analysis pose significant challenges for AI, especially when it comes to factors such as instrument calibration, chemical interactions, and environmental variability¹⁰.

The calibration curve is a graphical representation of the relationship between the concentration of a substance and its absorbance of light at a specific wavelength, following the Beer-Lambert law. Pharmaceutical analysis allows the quantification of an unknown sample by comparing its absorbance to a series of standards with known concentrations. This tool is essential for ensuring accurate dosing and assessing the purity and concentration of drugs in various formulations^7–9.

On the other hand, the dissolution test evaluates the rate at which an active drug is released from a solid dosage form, such as a tablet or capsule, into a dissolution medium. This test mimics the conditions in the gastrointestinal tract, providing critical information about how quickly a drug becomes available for absorption in the body. Dissolution testing is a key determinant of bioavailability, making it a critical component of both product development and post-market quality control. The dissolution behavior of a drug can impact its therapeutic efficacy and is tightly regulated by pharmacopeias and health authorities worldwide⁹.

With the advent of AI, the pharmaceutical industry is exploring new ways to streamline and improve various analytical processes. AI tools such as ChatGPT have shown potential for generating simulated data, assisting with data analysis, and predicting experimental outcomes. These tools can enhance efficiency, reduce costs, and provide valuable educational insights by allowing researchers and students to simulate laboratory processes before conducting actual experiments¹⁰.

Although AI has made significant strides in various scientific domains, its role in pharmaceutical quality control remains limited by the complexity and precision required in real laboratory environments. AI models may provide a reasonable approximation of experimental results but often fall short of replicating the intricacies of hands-on experiments^2,11. Factors such as instrument calibration, environmental conditions, and chemical interactions introduce variables that AI cannot fully account for, making real laboratory data indispensable¹².

This article investigates the capabilities and limitations of AI, specifically ChatGPT, in pharmaceutical analysis. By comparing AI-generated data for paracetamol dissolution tests and calibration curves with real experimental results, this study aimed to assess the utility of AI in approximating laboratory procedures. We explore the potential of AI tools in pharmaceutical technology, discuss their practical applications, and underscore the necessity of hands-on experimentation to achieve accurate, reliable data.

MATERIALS AND METHODS:

Materials:

For this study, paracetamol of pharmaceutical grade was sourced from Daar Al dawaa and used to construct calibration curves and conduct dissolution tests. Standard paracetamol solutions were prepared by dissolving precise quantities of paracetamol in deionized water to achieve concentrations ranging from 5 to 60 mg/L, which were then used for generating the calibration curve. The dissolution tests were performed in a phosphate buffer solution with a pH of 5.8, prepared according to standard laboratory protocols. Measurements of absorbance were conducted using a UV-Vis spectrophotometer (Shimadzu UV/Visible Scanning Spectrophotometer) set to a wavelength of 243 nm, while the dissolution tests were carried out using a dissolution apparatus (USP apparatus type 2) with a paddle system, maintaining the dissolution medium at 37°C and a paddle speed of 50rpm. Hypothetical data for the calibration curves and dissolution tests were generated using ChatGPT, developed by Open AI. Statistical analyses, including paired t-tests, were performed with R to compare real and AI-generated data. All other reagents and chemicals used were of analytical grade and obtained from Sigma Aldrich, including solvents, buffers, and laboratory consumables necessary for the experiments All solvents used were of HPLC grade.

Methods:

Calibration Curve Construction:

To construct the calibration curve, standard paracetamol solutions with known concentrations ranging from 5 to 60mg/L were prepared⁹. The absorbance values of these solutions were measured using a UV-Vis spectrophotometer at a wavelength of 243nm. The average absorbance was calculated from triplicate readings for each concentration and these values were used to generate a calibration curve⁸.

Dissolution Test:

In the dissolution test, paracetamol tablets were placed in 900mL of dissolution medium (phosphate buffer, pH 5.8) at 37°C with a paddle speed of 50rpm. Samples were collected at time intervals of 5, 10, 20, 30, 40, 50, and 60 min, and the percentage of paracetamol released was calculated in triplicate from UV absorbance readings using a calibration curve⁸.

AI-Assisted Analysis:

ChatGPT was tasked with generating hypothetical data for both the calibration curve and dissolution test based on the provided methodology. The AI-generated data were then compared with real experimental data to assess the accuracy and reliability of AI in simulating laboratory processes⁷.

RESULTS:

This section presents a comparison between the real experimental data and AI-generated data for both the calibration curve and dissolution tests. The accuracy and potential limitations of AI in simulating laboratory processes are highlighted by juxtaposing AI-predicted values with actual laboratory results^12,13. The calibration curve analysis shows how closely AI-generated data follow the expected linear trend but falls short in precision. Similarly, a comparison of dissolution tests demonstrated discrepancies in drug release kinetics, particularly in the early stages of the process³.

Calibration Curve: Real Data vs. AI-Generated Data:

Table 1 compares the real and AI-generated absorbance values for paracetamol at different concentrations.

Table 1: real and AI-generated calibration data for paracetamol

Concentration (mg/L)	Real Absorbance	AI Absorbance	Difference (AI - Real)	Squared Difference
10	0.201	0.108	-0.093	0.00864
20	0.411	0.216	-0.195	0.03803
30	0.579	0.318	-0.261	0.06812
40	0.705	0.424	-0.281	0.07896
50	0.813	0.532	-0.281	0.07904
60	0.921	0.640	-0.281	0.07904

Graphical Representation

Figure 1: Calibration curve of assay of paracetamol generated in the laboratory

Figure 2: Calibration curve of assay of paracetamol generated using ChatGPT:

· Real Calibration Curve: The experimental data exhibit a linear relationship between concentration and absorbance, as expected from the Beer-Lambert law.

· ChatGPT-Generated Calibration Curve: The AI-generated data also followed a linear trend, but the absorbance values were lower than the actual measurements. Despite this discrepancy, the general shape of the curve aligns with the expected behavior of paracetamol.

Dissolution Test: Real Data vs. AI-Generated Data:

The table below shows the real and AI-generated dissolution data for the paracetamol tablets.

Table 2: Dissolution data of the real and AI-generated

Time (min)	Real % Released	AI % Released	Difference (AI - Real)	Squared Difference
5	10.11	22	11.89	141.42
10	30.89	38	7.11	50.56
20	44.78	55	10.22	104.45
30	66.87	70	3.13	9.80
40	80.60	85	4.40	19.36
50	97.01	95	-2.01	4.04
60	100	99	-1.00	1.00

Graphical Representation

Figure 3: Dissolution profile of paracetamol tablet in phosphate buffer 5.8 (lab data)

Figure 4: Dissolution profile of paracetamol tablet in phosphate buffer 5.8 (ChatGPT data)

· Real Dissolution Profile: The experimental dissolution data showed a gradual release of paracetamol from the tablets, with 100% release achieved at 60 min.

· AI-Generated Dissolution Profile: The AI-generated data also showed a progressive increase in drug release over time, but the initial release rate was significantly higher. AI overestimates the percentage of drugs released in the earlier stages, although it closely matches the real data at later time points.

Statistical calculations:

I. Mean Squared Error (MSE):

· Dissolution Test MSE:

MSE = (Sum of squared differences) / n

MSE = (141.42 + 50.56 + 104.45 + 9.80 + 19.36 + 4.04 + 1.00) / 7

MSE ≈ 47.80

· Calibration Curve MSE:

MSE = (Sum of squared differences) / n

MSE = (0.00864 + 0.03803 + 0.06812 + 0.07896 + 0.07904 + 0.07904) / 6

MSE ≈ 0.0583

II. Percent Error

· Dissolution Test percentage Error:

Percent Error = (Mean of absolute differences / Mean of real values) × 100 Percent Error = (7.39 / 61.18) × 100 ≈ 12.08%

· Calibration Curve Percent Error:

Percent Error = (Mean of absolute differences / Mean of real values) × 100

Percent Error = (0.232 / 0.605) × 100 ≈ 38.35%

III. Correlation Coefficient (R²)

· Dissolution Test R²:

Using a simplified formula for the correlation coefficient, R² for dissolution was ≈ 0.988, showing a strong correlation.

· Calibration Curve R²:

The R² for calibration was ≈ 0.998, indicating an almost perfect correlation.

DISCUSSION:

Calibration Curve Analysis:

The AI-generated calibration curve followed a similar trend to the real experimental data, demonstrating the ability of AI to approximate the relationship between concentration and absorbance. However, the absorbance values of AI are consistently lower than the real values, which could be because AI does not account for instrument-specific factors such as baseline absorbance, light scattering, or impurities in the solution. Despite these discrepancies, the linearity of the AI-generated curve remains intact, making it useful for theoretical predictions and educational purposes^14,15.

In a real laboratory setting, precise calibration of the instruments is crucial for obtaining accurate results. The deviations in the AI-generated absorbance values highlight the limitations of using AI models for accurate experimental predictions. However, AI can still play a valuable role in simulating expected trends and helping students and researchers understand the underlying principles before conducting actual experiments¹⁶.

Dissolution test analysis:

The AI-generated dissolution data showed a faster initial release of paracetamol compared to the real data. This could be due to the inability of AI to fully account for factors such as tablet disintegration, stirring conditions, and the interaction between the drug and dissolution medium. In contrast, the real experimental data showed a more gradual release, which aligns with the expected behavior of paracetamol tablets.

Despite the differences in the early stages of the dissolution process, the AI-generated data closely matched the real data at later time points (50 and 60 min). This suggests that although AI can provide an approximate model of drug dissolution, it cannot accurately predict the complex kinetics of drug release without considering the numerous variables present under real laboratory conditions¹⁷.

To assess the statistical significance of the differences between the real and AI-generated data, we conducted a paired t-test for both the calibration curve and dissolution test data. For the calibration curve, the differences between the real and AI-generated absorbance values were calculated, yielding a mean difference of -0.222 and a standard deviation of 0.085. The resulting t-statistic was approximately -4.94, with five degrees of freedom, indicating a highly significant difference with a p-value less than 0.01. Similarly, for the dissolution test, differences in the percentage of paracetamol released were analyzed, and the paired t-test was applied. The significant t-values and corresponding p-values for both datasets highlight that while AI-generated data can approximate trends, there are substantial discrepancies compared to real experimental results, underscoring the limitations of AI in replicating precise laboratory conditions.

Importance of real laboratory data:

The comparison between the real and AI-generated data underscores the importance of actual laboratory experimentation. Although AI can simulate general trends and provide useful insights for theoretical understanding, it cannot replicate the precision and accuracy of real-world measurements. Factors such as instrument calibration, environmental conditions, and human oversight play critical roles in obtaining reliable data that AI models cannot fully capture. As pharmaceutical technology continues to evolve, the integration of AI tools such as ChatGPT presents new opportunities and challenges for the industry. AI's potential of AI to streamline data analysis, simulate laboratory processes, and predict outcomes offers significant advantages for pharmaceutical technology, especially in the realms of quality control and drug formulation. In this study, AI-assisted methods were employed to analyze paracetamol dissolution tests and construct calibration curves, demonstrating the capability of the technology to approximate experimental results. However, the discrepancies observed between the AI-generated and real experimental data underscore the limitations of AI in replicating the precise conditions and variables inherent to laboratory settings. The role of AI in pharmaceutical technology is thus seen as complementary rather than substitutive, providing valuable theoretical insights and aiding in experimental planning, but requiring the accuracy and rigor of hands-on laboratory work for definitive results ¹⁶. This dual approach, which combines AI tools with traditional experimental techniques, represents a promising avenue for advancing pharmaceutical technology and enhancing the efficiency of drug development and quality control processes. The integration of AI into pharmaceutical technology, particularly in generating dissolution models and calibration curves, offers significant potential for both research and education. Although AI-generated data may lack the precision and accuracy required for regulatory submissions, its ability to simulate general trends makes it an invaluable tool in drug development and educational settings ¹⁸. For new drugs, AI can be used to model dissolution profiles and predict calibration curves during the early stages of development, enabling researchers to assess potential behaviors without the immediate need for extensive hands-on experiments. This saves time and resources and accelerates the decision-making process in the development pipeline. Furthermore, AI-powered simulations can be particularly useful in academic environments where students can use virtual laboratories to visualize the principles of dissolution testing and analytical methods. By reducing dependency on costly laboratory equipment and consumables, AI helps educators provide more accessible and flexible training opportunities. Although current AI technology cannot fully replicate the complexity of real laboratory experiments, its role in supplementing hands-on work is undeniable. As AI continues to evolve, its integration into pharmaceutical technology will play a pivotal role in optimizing research, enhancing education, and streamlining drug development processes, ultimately driving innovation and efficiency across industries.

CONCLUSION:

This study demonstrates the potential of AI tools such as ChatGPT in assisting with the analysis of paracetamol dissolution tests and calibration curves. Although AI-generated data can provide valuable theoretical insights and approximate trends, they cannot replace the accuracy and reliability of real laboratory experiments. The discrepancies between the real and AI-generated data highlight the need for hands-on experimentation, particularly in pharmaceutical analysis, where precision and reproducibility are essential.

AI tools can be used to supplement laboratory work by helping researchers and students visualize expected outcomes and plan experiments. However, they should not be relied upon as substitutes for real-world data. The combination of AI-generated models and empirical laboratory work provides the best approach for understanding complex scientific phenomena.

ACKNOWLEDGMENTS:

We would like to extend our sincere gratitude to Jadara University for providing access to their laboratory facilities and instruments, which were instrumental in conducting the analyses in this study.

FUND:

We also acknowledge the financial support provided by Jadara University, which made this research possible.

CONFLICT OF INTEREST:

The authors declare that there are no conflicts of interest related to this study.

ABBRIVIATIONS:

AI: Artificial Intelligence

API: Active Pharmaceutical Ingredient

UV-Vis: Ultraviolet-Visible

pH:Potential of Hydrogen (or Hydrogen Ion Concentration)

rpm: Round Per Minute

mg/L: Milligrams Per Liter

REFERENCES:

1. Kumar S, Malviya R, Sharma PK. Solid dispersion: Pharmaceutical technology for the improvement of various characteristics of active pharmaceutical ingredients. African Journal of Basic and Applied Sciences. 2011; 3(4): 116–25.

2. Malkawi R. Revolutionizing drug delivery innovation: leveraging AI-driven chatbots for enhanced efficiency. International Journal of Applied Pharmaceutics. 2024; 16(2): 52–6.

3. Malkawi R, Malkawi WI, Al-Mahmoud Y, Tawalbeh J. Current trends on solid dispersions: past, present, and future. Adv Pharmacol Pharm Sci. 2022; 14(4): 1–17.

4. Jarwan B, Tawalbeh J, Malkawi R. Assessment of phenol and antioxidant content of olive varieties and their potential health benefits for colon health. The Scientific World Journal. 2023; 5: 41–52.

5. Malkawi R, Jarwan B, Tawalbeh J. Assessing the awareness, attitude, and knowledge of senior pharmacy students in Jordanian universities regarding antibiotic use and resistance: a study of the medical curriculum. J Pharm Res Int. 2023 Mar 25; 35(5):41–52.

6. Zhao A, Wu Y. Future implications of ChatGPT in pharmaceutical industry: drug discovery and development. Front Pharmacol. 2023Jun 17; 14(5): 1–14.

7. Ji Y, Lemberg M, Prudic A, Paus R, Sadowski G. Modeling and analysis of dissolution of paracetamol/Eudragit® formulations. Chemical Engineering Research and Design. 2017 May 1; 121(8): 22–31.

8. Abebe K, Beressa TB, Yimer BT. In-vitro evaluations of quality control parameters of paracetamol tablets marketed in gondar city, northwest ethiopia. Drug Healthc Patient Saf. 2020 Dec 21; 12(10): 273–9.

9. Menon D, Shilpa K. “Chatting with ChatGPT”: Analyzing the factors influencing users’ intention to use the open AI’s ChatGPT using the UTAUT model. Heliyon. 2023 Nov; 9(11): 209–62.

10. Oh AL, Hassali MA, Al-Haddad MS, Sulaiman SAS, Shafie AA, Awaisu A. Public knowledge and attitudes towards antibiotic usage: a cross-sectional study among the general public in the state of Penang, Malaysia. The Journal of Infection in Developing Countries. 2010 Nov 9; 5(5): 338–47.

11. Juhi A, Pipil N, Santra S, Mondal S, Behera JK, Mondal H, et al. The capability of ChatGPT in predicting and explaining common drug-drug interactions. Cureus. 2023 Mar; 15(3): 1–7.

12. Malone DC, Abarca J, Skrepnek GH, Murphy JE, Armstrong EP, Grizzle AJ, et al. Pharmacist workload and pharmacy characteristics associated with the dispensing of potentially clinically important drug-drug interactions. Med Care. 2007 May; 45(5): 456–62.

13. Heck TG. What artificial intelligence knows about 70 kDa heat shock proteins, and how we will face this ChatGPT era. Cell Stress Chaperones. 2023 May; 28(3): 225–9.

14. Parmar A, Lakshminarayanan R, Iyer A, Mayandi V, Leng Goh ET, Lloyd DG, et al. Synthesis and Biological Evaluation of a Teixobactin Analogue. Org Lett. 2015 Dec 18; 17(24): 6182–5.

15. Sallam M. ChatGPT utility in healthcare education, research, and practice: systematic review on the promising perspectives and valid concerns. Healthcare . 2023 Mar; 11(6): 1–20.

16. Fonseca-Berzal C, Palmeiro-Roldán R, Escario J, Torrado S, Arán V, Torrado-Santiago S, et al. Novel solid dispersions of benznidazole: Preparation, dissolution profile and biological evaluation as alternative antichagasic drug delivery system. Exp Parasitol. 2015 Feb 1; 149: 84–91.

17. Makawi R, Al-olimat S, Tawalbeh J. Unlocking the potential: enhancing solubility and bioavailability of acyclovir through solid dispersion formulations. International Journal of Applied Pharmaceutics. 2024 Sep 7; 15(5): 111–8.

Received on 15.09.2024 Revised on 06.01.2025

Accepted on 21.03.2025 Published on 02.05.2025

Available online from May 07, 2025

Research J. Pharmacy and Technology. 2025;18(5):2269-2274.

DOI: 10.52711/0974-360X.2025.00325

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License. Creative Commons License.