Estudo de Estimativa de Medicamentos - Farmacologia I (2024)

Prévia do material em texto

Sakurand AlZakri BMC Chemistry (2024) 18:17 https://doi.org/10.1186/s13065-024-01126-1RESEARCH Open Access© The Author(s) 2024. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. The Creative Commons Public Domain Dedication waiver (http:// creat iveco mmons. org/ publi cdoma in/ zero/1. 0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.BMC ChemistryThe effectiveness ofmultivariate andunivariate spectrophotometric techniques fortheconcurrent estimation ofornidazole andciprofloxacin HCl intablet formulation andspiked serum: estimating greenness andwhiteness profileAmir A. Sakur1* and Duaa Al Zakri1 Abstract In this manuscript, the effectiveness of multivariate and univariate tools in conjunction with spectrophotometric techniques was evaluated for the concurrent analysis of ciprofloxacin (CI) and ornidazole (OR) in prepared mixtures, tablets, and human serum. The artificial neural network was chosen as the multivariate Technique. Bayesian regulariza-tion (trainbr) and Levenberg–Marquardt algorithms (trainlm), were constructed and trained using feed-forward back-propagation learning. The optimal logarithm was determined based on mean recovery, mean square error of predic-tion (MSEP), relative root mean square error of prediction (RRMSEP), and bias-corrected MSEP (BCMSEP) scores. Trainbr outperformed trainlm, yielding a mean recovery of 100.05% for CI and 99.84% for OR, making it the preferred algo-rithm. Fourier self-deconvolution and mean-centering transforms were chosen as the univariate Techniques. Fourier self-deconvolution was applied to the zero-order spectra of ciprofloxacin and ornidazole by electing an appropriate full width at half maximum, enhancing peak resolution at 380.1 nm and 314.2 nm for CI and OR, respectively. Mean centering transform was applied to CI and OR ratio spectra to eliminate constant signals, enabling accurate quantifi-cation of CI and OR at 272.0 nm and 306.2 nm, respectively. The introduced approaches were optimized and validated for precise CI and OR analysis, with statistical comparison against the HPLC method revealing no notable differ-ences. The sustainability of these approaches was confirmed through the green certificate (modified eco-scale), AGP, and whiteness-evaluation tool, corroborating their ecological viability.Keywords Artificial intelligence, Bayesian Regularization, Fourier self-deconvolution, Green certificate, Levenberg–Marquardt, Mean centeringIntroductionvagin*l dysbiosis poses a significant challenge in obstet-rics and gynecology, owning a prevalence of 10–30% in women worldwide. vagin*l dysbiosis is common over pregnancy and several studies have noted a relation between bacterial vaginosis and premature birth in addi-tion to miscarriage and low birth weight [1, 2]. Recent *Correspondence:Amir A. Sakurprofsakur@gmail.com1 Analytical and Food Chemistry Dept., Faculty of Pharmacy, Aleppo University, Aleppo, SyriaContent courtesy of Springer Nature, terms of use apply. Rights reserved.Page 2 of 16Sakurand AlZakri BMC Chemistry (2024) 18:17research highlights the therapeutic potential of a cipro-floxacin HCl and ornidazole combination for manag-ing bacterial vaginosis and vagin*l dysplasia associated with mixed anaerobic and aerobic bacterial presence [3]. Another study underscores the clinical efficacy and tol-erability of an antimicrobial-antiprotozoal combination comprising ornidazole and ciprofloxacin in postoperative complications, cystitis, and overactive bladder treatment [4].Ornidazole (OR), Fig. 1, exhibits bactericidal activity against anaerobic bacterial infections such as amoebae and trichom*onas, disrupting microbial DNA synthesis [5, 6]. Ciprofloxacin HCl (CI), Fig.1, is a broad-spectrum antibacterial agent effective against respiratory, urinary, and skin infections, inhibiting bacterial DNA replication [7, 8].Previous literature has reported the estimation of cip-rofloxacin and ornidazole combination in tablet form using HPLC [9–12] and spectroscopy [13–16]. Concur-rent estimation in spiked serum has been documented using TLC [17] and HPLC [18] only.Multivariate technique represented by an artificial neu-ral network (ANN) is of significant importance. ANN eliminates the error resulting from employing a single wavelength regression and is therefore robust. Also, the various concentration ratios covered by ANN are highly broad and can include future combinations of ratios created by pharmaceutical corporations. Univariate techniques like Fourier self-deconvolution, and mean centering are distinguished by simplicity in resolving the interfering drugs without requiring complicated math-ematical procedures or affecting the signal-to-noise ratio.This study pioneers the employment of artificial intel-ligence via two logarithmic methods (trainbr, trainlm), Fourier self-deconvolution, and mean centering in spec-trophotometric analysis for concurrent quantification of CI and OR in pharmaceutical formulations and human serum without complex sample preparation or separa-tion. The strengths and limitations of these approaches are debated comprehensively, highlighting their perfor-mance. Furthermore, a statistical comparative study was conducted between the introduced approaches and the HPLC approaches to assure efficacy.In alignment with the role analytical laboratories play in environmental protection by monitoring pollutants, this work embraces the principles of green chemistry, emphasizing operator safety, responsible solvent usage, waste reduction, and energy efficiency [19, 20]. This endeavor thus provides eco-friendly, intelligent, and straightforward methodologies for routine CI and OR analysis.TheoryArtificial neural network (ANN)An Artificial Neural Network (ANN) is an effective model that simulates the functioning of the human brain’s neural networks. Neurons, the basic units of ANN, struc-turally resemble human nerve cells. Training an ANN involves adjusting its weights and biases to attain a spe-cific objective as illustrated in Fig. 2 [21]. In this study, a feed-forward network trained with backpropagation [22] was used. Different learning algorithms exist for neural networks, including Levenberg–Marquardt back-propagation, Bayesian regularization backpropagation, Resilient backpropagation, and One-step secant back-propagation, among others. For this study, we employed two algorithms: Bayesian regularization backpropagation Fig. 1 UV spectra and chemical structure of CI and ORFig. 2 Network training methodContent courtesy of Springer Nature, terms of use apply. Rights reserved.Page 3 of 16Sakurand AlZakri BMC Chemistry (2024) 18:17 and Levenberg–Marquardt backpropagation. Bayesian regularization reduces squared errors and weight combi-nations, ensuring a generalized network. It is robust and does not demand validation [23]. Levenberg–Marquardt, introduced by Donald Marquardt, addresses nonlinear least squares issues and enhances second-order training speed without necessitating the computation of the Hes-sianmatrix, as in Newton’s algorithm [24].Fourier self‑deconvolution (FSD)This function in the spectra software is an advanced combination of the Fourier self-deconvolution tech-nique (FSD) proposed by Kauppinen et al. and the Finite Impulse Response Operator edited by Jones and Shimokoshi [25]. Given E(ν) as the actual spectrum, M(ν) as the measured spectrum, and G(ν) as the spectrum altered by instrumental measurement, the deconvoluted spectrum is expressed through a Fourier transformation (F) as follows:Here, D(x, L) is the tapering function, and exp{πσ x} represents the Lorentzian curve’s half-width of G(ν).To optimize computation time, an algorithm that com-bines the FSD and FIRO techniques is utilized, where the FIRO technique includes extending the exp{ πσ x}expres-sion in the previous equation on the spectrum area for diverse types of tapering functions. The deconvolution filter applied to the spectrum takes the form:Here, D(x, L) and D′(v, FL) are constants related to the Bessel function, and L and FL are automatically deter-mined values. The FSD algorithm built within the Spectra Manager software is a powerful technique for resolving overlapping peaks, achieved by applying an appropriate full width at half maximum (FWHM) to the spectrum, producing a deconvoluted spectrum with zero-crossing points (λzero). The concentration of each component is detected then by the constructed linear equation between the signal at λzero and the analogous concentrations.Mean centering (MC)Mean centering is a data transformation technique via MATLAB® designed to handle highly overlapped signals by eliminating redundant signals while preserving vari-able signals [26].Considering a vector with 3-dimensional:E(v) = F{D(x, L)exp{πσ |x|}}∗M(v)F{D(x, L)exp{πσ |x|}}∗ D′(v, FL)The previous column is then mean-centered by deducting the average of the three numbers.Utilizing x =444MC(x) = x − x =363−444 =−12−1In a mixture consisting of X and Y components, the absorbance (A) is expressed as:Where (a) represents the absorptivity factor, and (C) represents the component’s concentration.When dividing Eq.(1) by component Y spectrum, we obtain:Utilizing MATLAB®, mean centering is applied to the ratio spectra, resulting in:The constant’s mean centering value is zero. Equa-tion (3) illustrates that after mean centering, the mix-ture signal is solely related to component X.ExperimentalInstruments andsoftwareThe JASCO V-650 spectrophotometer was employed for D0 spectrum scanning in the range of (200–400) nm. Spectra Manager® software, version 2 by JASCO Corporation, was used for Fourier self-decon-volution. MATLAB® 2021a (version 8.6) was utilized for the ANN and mean-centering approaches.Materials andsolvent– Ciprofloxacin HCl and Ornidazole were obtained from Chongqing Chemdad CO., Ltd., CHINA, with purities of 99.35 ± 0.73 and 99.16 ± 0.62, respec-tively.– Methanol was procured from Panreac, Spain.– Cifran-OZ® film-coated tablets, each labeled to contain 500 mg of ornidazole and ciprofloxacin HCl, were manufactured in India by Sun Pharma Laboratories Ltd., with batch no. SXC0964A.x =363(1)AM = ax Cx + ay Cy(2)AM/ay = ax Cx/ay + Cy(3)MC(AM/ay)= MC(ax Cx/ay)Content courtesy of Springer Nature, terms of use apply. Rights reserved.Page 4 of 16Sakurand AlZakri BMC Chemistry (2024) 18:17Standard solutions– Stock standard solutions of 1000 μg/ml were pre-pared for both CI and OR in methanol.– Working solutions of 50 μg/ml were prepared for both CI and OR in methanol.Preparing tablet sampleTen Cifran-OZ® tablets were weighed accurately, and crushed, and an amount equivalent to 10mg of each CI and OR was dissolved in a 50 ml standard flask using 20ml of methanol. The solution was sonicated for 5min, filtered into a 50ml standard flask, and topped up with methanol. Appropriate dilution was performed to pre-pare the sample solution.Preparing serum sampleThe serum sample was human serum collected from three healthy volunteers, then treated as follows: 0.1ml of serum sample was mixed with various concentrations of CI and OR in centrifugation tubes. The volume was adjusted to 5 ml using methanol, and the mixture was vortexed for 2min and then centrifuged at 5000rpm for 10min. The centrifuged serum sample was filtered utiliz-ing a 0.22µm Millipore filter and measured against a pro-duced blank under the aforementioned conditions except for the existence of the drugs.Procedures andcalibrationA series of methanolic solutions were prepared in the laboratory, spanning concentrations from 2 to 15 μg/ml for CI and 3–25μg/ml for OR. These solutions were recorded against methanol and saved in the spectra software.Artificial neural network (ANN)For trainbr, thirty-two mixtures containing varying ratios of CI and OR, within the concentration ranges of 2–15μg/ml for CI and 3–25μg/ml for OR, were prepared for the training set (Table1). Additionally, a test set com-prising 10 mixtures of CI and OR was prepared to evalu-ate the trained artificial networks. For trainlm, thirty mixtures of CI and OR were utilized as the training test, 6 mixtures as the validation set, and an additional 6 mix-tures as the test set.Fourier self‑deconvolution methodThe recorded spectra of CI and OR were subjected to Fou-rier self-deconvolution (FSD) using a full-width half-max-imum (FWHM) of 70. A calibration plot was established between the signal values at 280.1nm for CI and 314.2nm for OR, corresponding to their respective concentrations.Mean centering methodThe spectra of CI and OR were divided by the spectra of CI (10μg/ml) and OR (20μg/ml), respectively. The result-ing ratio spectra were subjected to mean centering. Lin-ear equations were developed between the signals at 272.0nm for CI and 306.2nm for OR, and their respective concentrations.Results anddiscussionUpon analyzing Fig. 1, it is apparent that the D0- spec-tra associated with CI and OR overlap in a manner that hampers the independent quantification of each Table 1 Concentrations of training mixtures in μg/mlMix no. CI μg/ml OR μg/ml1 2.0 3.02 3.0 5.03 5.0 10.04 8.0 15.05 2.5 10.06 2.0 5.07 5.0 15.08 3.0 3.09 5.0 5.010 8.0 10.011 10.0 3.012 15.0 0.013 3.0 10.014 2.0 10.015 3.0 15.016 5.0 20.017 8.0 25.018 3.0 10.019 5.0 15.020 2.0 20.021 3.0 25.022 5.0 15.023 8.0 15.024 10.0 5.025 2.0 15.026 3.0 10.028 5.0 15.028 8.0 20.029 2.0 25.030 3.0 3.031 5.0 25.032 8.0 3.0Content courtesy of Springer Nature, terms of use apply. Rights reserved.Page 5 of 16Sakurand AlZakri BMC Chemistry (2024) 18:17 compound. As a solution, this study introduces three approaches to permit the accurate quantification of indi-vidual compounds. The potential of ANN was explored to resolve the CI and OR overlap within the UV range (279.0–316.0) nm at a 0.5 nm interval. The ANN input consisted of an absorbance matrix (75 × 32), while the output matrix (32 × 2) represents the concentrations for CI and OR. The Fourier Self-Deconvolution (FSD) tech-nique’s efficiency was assessed through variations in the Full Width at Half Maximum (FWHM) for CI and OR spectra within the UV range (200.0–400.0) nm using a 0.1nm interval. Additionally, the Mean Centering (MC) approach was applied to CI and OR ratio spectra within the UV range (240.0–290.0) nm for CI and (260.0–340.0) nm for OR using a 0.1nm interval.ANNTo ensure the correctness of the ANN training process and minimize errors, it was crucial to carefully select noise-free and reliable absorbance values for input. In this context, signal points within the range of (279.0–316.0) nm were chosen as the network’s inputs, while other spectral points were disregarded. For the ANN training,two algorithms, trainbr, and trainlm, were employed, and three network configurations were considered for each component, consisting of (1 hidden layer, 1 neuron), (1 hidden layer, 2 neurons), and (2 hidden layers, 2 neu-rons). Activation functions TANSIG and PURELIN were applied in the hidden and output layers, respectively. The output layer of all networks contained only 2 neurons to accommodate the two compounds in the mixtures. It has been determined that training networks with one hidden layer and two hidden layers are sufficient to address the spectral interference of CI and OR. Further hidden layers could potentially exacerbate the issue of overfitting [27]. The performance of each network was evaluated using various parameters, including mean recovery%, mean square error of prediction (MSEP), relative root mean square error of prediction (RRMSEP), and bias-corrected MSEP (BCMSEP) [28].The formulas utilized for the aforementioned param-eters are as follows:MSEP =∑ni=1(c′ − c)2nRRMSEP =100c√MSEPBCMSEP =∑ni=1(c′ − c)2 −[∑ni=1(c′ − c)2/n]n− 1Here, c’: represents the predicted concentration, c: rep-resents the true concentration, c̅: is the average of true concentrations, and n is the sample number.Notably, all trained networks performed well except the (1 hidden layer, 1 neuron) configuration, which exhib-ited suboptimal outcomes (Figs. 3, 4). Trainbr demon-strated fewer errors in performance with superior results in terms of MSEP, BCMSEP, and RRMSEP compared to trainlm. Also, as noticed from Figs. 5, 6, 7, 8 the slope value of 1 in trainbr diagrams indicates the excellent performance of the trained net, unlike trainlm, which exhibits a slope value of less than 1 in test diagrams of CI. While there was no notable difference between trainbr with (1 hidden layer, 2 neurons) and trainbr with (2 hidden layer, 2 neurons), the former was chosen for the simultaneous detection of CI and OR in tablet and serum samples. As depicted in Fig. 5 the training and test diagrams of trainbr (1 layer, 2 neurons) for CI and OR reveal an excellent R-value. Furthermore, the mean recovery% and RMSE scores provided in Table2 validate the algorithm’s exceptional performance. These satisfac-tory results were achieved within 81 epochs, where the gradual decrease of training and test lines confirms the absence of overfitting as indicated in Fig.9.Fourier self‑deconvolution (FSD)The FSD technique is a significant processing tool that enhances the resolution of overlapping spectra by adjust-ing the peak widths. This adjustment allows for achieving zero intersection points for each spectrum, thereby ena-bling the quantification of individual compounds within the drug mixture. Unlike derivative transformation-based methods [29–31], FSD maintains the signal-to-noise ratio and avoids introducing additional noise. To determine the appropriate Full Width at Half Maximum (FWHM) for deconvolution, various FWHM values (10, 30, 50, 70) were applied to the individual spectra of CI and OR within the UV range (200–400) nm using a 0.1 nm interval. The optimal FWHM for deconvolution was determined to be 70, as it allowed for the measurement of each drug without intervention from the other com-pound. This was achieved at 280.1nm for CI (Fig.10) and 314.2nm for OR (Fig.11). The following linear equations are utilized to estimate the concentration of each drug at its elected wavelength: for CI Y = 0.2732 X + 0.0031, for OR Y = 0.0778 X + 0.002.Mean centering(MC)This tool of data transformation utilizing MATLAB® has the feature of picking the felicitous wavelength for measurement in an expansive range of wavelengths without being restricted to choosing zero points. To Content courtesy of Springer Nature, terms of use apply. Rights reserved.Page 6 of 16Sakurand AlZakri BMC Chemistry (2024) 18:17Fig. 3 MSEP (a), RRMSEP (b), and BCMSEP (c) results for test samples of CI and ORFig. 4 Mean recovery% results for test samples of CI and ORContent courtesy of Springer Nature, terms of use apply. Rights reserved.Page 7 of 16Sakurand AlZakri BMC Chemistry (2024) 18:17 apply this method two criteria were tested: the divi-sor concentration and wavelength range to perform mean centering. The best divisor of CI and OR which accords the best sensitivity and precision were 10 μg/ml of CI and 20 μg/ml of OR, also many wavelength ranges were tried to elect the best range that fulfills the best recovery% scores of CI and OR from the pro-duced mixes as in Table 3, and the elected ranges to accomplish the estimating operation of CI and OR were (240.0–290.0) nm for CI (Fig. 12a) and (260.0–340.0) nm for OR (Fig. 12b). After applying the mean Fig. 5 Training, and test diagrams of trainbr (1 layer, 2 neurons) for CI and ORFig. 6 Training, and test diagrams of trainbr (2 layers, 2 neurons) for CI and ORContent courtesy of Springer Nature, terms of use apply. Rights reserved.Page 8 of 16Sakurand AlZakri BMC Chemistry (2024) 18:17centering function on the produced ratio spectra of CI and OR through the elected wavelength ranges, the λmax at 272.0 nm was elected to detect CI using the following equation Y = 0.2091 X + 0.0026, and the λmax at 306.2nm was elected to detect OR utilizing the fol-lowing equation Y = 0.0571 X + 0.0042.Fig. 7 Training, validation, and test diagrams of trainlm (1 layer, 2 neurons) for CI and ORFig. 8 Training, validation, and test diagrams of trainlm (2 layers, 2 neurons) for CI and ORContent courtesy of Springer Nature, terms of use apply. Rights reserved.Page 9 of 16Sakurand AlZakri BMC Chemistry (2024) 18:17 Strengths points andlimitations oftheelaborated toolsDespite the satisfactory and convergent results attained through the elaborating of the three tools for the syn-chronous quantification of CI and OR, these tools differ from each other by several points that are epitomized in Table4 to display the strengths points, and limitations of each tool.Confirmation ofenvironmental sustainabilityIn the current era, the development of analytical meth-ods must prioritize environmental considerations, aim-ing to uphold ecological balance, minimize waste, reduce energy consumption, and maintain economic viability. Previous research has introduced various tools for eval-uating the sustainability of methods. In this study, the Table 2 Prediction outcome of test mixes utilizing trainbr (1 layer, 2 neurons)Test mixture Real μg/ml Expected μg/ml Recovery%CI OR CI OR CI OR1 0.0 10.0 0.01 9.89 – 98.932 10.0 15.0 9.99 15.28 99.87 101.843 5.0 10.5 4.93 10.64 98.58 101.304 6.0 12.0 5.91 11.73 98.47 97.765 8.0 8.0 8.19 7.90 102.42 98.706 10.0 20.0 10.18 19.86 101.81 99.297 12.0 4.0 12.29 3.92 102.41 98.028 12.0 6.0 11.89 6.07 99.09 101.259 15.0 5.0 14.87 5.13 99.11 102.5410 10.0 10.0 9.87 9.88 98.73 98.79Recovery% 100.05 99.84RMSE 0.073 0.077Fig. 9 Best training performance of trainbr (1 layer, 2 neurons) for CI and OR predictions in the produced mixesContent courtesy of Springer Nature, terms of use apply. Rights reserved.Page 10 of 16Sakurand AlZakri BMC Chemistry (2024) 18:17sustainability of the proposed approaches was affirmed through the creation of greenness and whiteness profiles using the following methodologies:Green certificate (modified Eco‑scale)A new amendment of the eco-scale tool addresses diverse analytical evaluation aspects like reagent volume, hazard, power intake, occupational danger, and generated quan-tity of wastes in the analysis. This tool relies on a pen-alty points system and a specific mathematical process to classify the analytical approaches into seven classes according to the final penalty points score; Fig. 13. The penalty points of utilized solvent volume and generated waste are estimated via the following equation:where a = 0.61 ± 0.05, b = 0.31 ± 0.02for solvent con-sumption and a = 1.50 ± 0.08, b = 0.40 ± 0.02 for produced waste. The obtained penalty points for solvent consump-tion should be multiplied by the penalty points of hazard [32].y = a× xbFig. 10 Applying FSD on the UV-spectra of CI(____) 2–15 μg/ml and OR(----) 10 μg/ml using FWHM: 70, displaying zero-crossing point for CI estimationFig. 11 Applying FSD on the UV spectra of OR(____) 3–25 μg/ml and CI(----) 10 μg/ml using FWHM: 70, displaying zero-crossing point for OR estimationContent courtesy of Springer Nature, terms of use apply. Rights reserved.Page 11 of 16Sakurand AlZakri BMC Chemistry (2024) 18:17 AGP ToolThis tool evaluates the ecological impact of the proce-dure based on five criteria: health impact, procedural safety, energy consumption, waste production, and eco-logical hazards. The evaluation is represented through a pentagram divided into five segments, as illustrated in Fig. 13. The color-coded segments, ranging from green to yellow and red, indicate the varying degrees of eco-logical impact–high, moderate, and low. Health and safety impacts are estimated based on the red and blue segments of the solvent’s NFPA pictogram, respectively. Waste and ecological hazards are evaluated by consider-ing generated waste during the operation, while energy consumption is estimated based on device usage [33].Whiteness profileThe whiteness profile of the tools was assessed using the RGB 12 algorithm. This method involved populating an Excel template worksheet that comprises three distinct tables: red, green, and blue. Each table corresponds to specific criteria. The green table summarizes factors such as trueness, precision, detection, and quantification limit. The red table focuses on safety, toxicity, and energy con-sumption, while the blue table addresses cost, speed, and simplicity. The amalgamation of these colors produces white, symbolizing the extent of brightness associated with the method, as illustrated in Fig. 13. By assigning a score from 0 to 100 to each criterion in the Table (0 Table 3 Effect of wavelength ranges on the recovery results of the prepared mixtures of CI and ORa The elected wavelength ranges for mean centering transformsWavelength range Linear equation Recovery% Drugs concentration (μg/ml) CI: OR10:5 5:15 3:10CI240–290 nma Y = 0.2091X + 0.0026 99.42 98.92 98.10230–300 nm Y = 0.2630X + 0.0030 95.70 96.61 96.41260–290 nm Y = 0.1206X + 0.0015 96.07 95.18 92.98240–325 nm Y = 0.2943X + 0.0037 95.96 92.08 92.14250–300 nm Y = 0.2118X + 0.0028 97.03 95.78 96.82OR260–340 nma Y = 0.0571X + 0.0042 98.13 99.12 101.09290–330 nm Y = 0.0231X + 0.0030 96.27 93.33 97.06240–340 nm Y = 0.0609X + 0.0040 104.21 103.94 104.80260–330 nm Y = 0.0581X + 0.0041 97.02 97.15 103.06280–340 nm Y = 0.0401X + 0.0037 98.37 93.71 97.03Fig. 12 Applying the mean centering function on the ratio spectra of CI (a) and OR (b)Content courtesy of Springer Nature, terms of use apply. Rights reserved.Page 12 of 16Sakurand AlZakri BMC Chemistry (2024) 18:17representing the worst result and 100 signifying a well-fitted method), the final whiteness score, ranging from 0 to 100, is automatically generated [34].ValidationValidation ofANNThe self-validating nature of ANN is a key advantage, and several parameters were computed to assess predic-tion accuracy [28]. Prediction accuracy was evaluated via MSEP and RRMSEP, while prediction precision was eval-uated using BCMSEP. The computed parameters yielded exceptional scores, as summarized in Table5.Validation offourier self‑deconvolution andmean centering methodsThe validation of the FSD and MC approaches adhered to ICH guidelines [35] and encompassed the following aspects:LinearityCalibration plots were established for both CI and OR by plotting the signal against their corresponding concen-trations. Remarkable linearity was achieved with an R2 value exceeding 0.999, as detailed in Table6.Table 4 The strengths points and limitations of the elaborated ToolsTool Benefits LimitationsANN • Eliminates the error reverting to employing a single wavelength regression such as in univariate UV approaches• Does not demand a monotonous validation process• Can be trained on diverse ratios of mixes• Needs many tests to obtain accurate prediction outcomes, which needs to adjust many parameters like transfer functions, layers, neuron number, goal, and the learning rate• The improper adjustment of one of the previous parameters may lead to learning errors or the occurrence of overfitting issuesFSD • Enhances the overlaid spectra resolution• Does not impact the signal/noise ratio• Doesn’t require a special program or (cos, sin) transformation for per-forming like in Discrete Fourier Transform• Crucial measurements of the signals at the selected wavelength• Influenced by the increment of wavelengthMC • The mean-centered signals are measured at the maximum points, for higher sensitivity• Does not impact the signal/noise ratio• The computed arithmetic mean is greatly affected by skewed data• Needs to test the best divisor concentration and the best wavelength range for the mean centering processFig. 13 Evaluation of the environmental sustainability of the developed approaches using green certificate, AGP, and whiteness assessmentTable 5 Construction and Validation Parameters of the ANN approacha Calculated for CI and OR at concentrations of (5µg/mL, 6µg/mL, 8µg/mL, 10µg/mL, 12µg/mL, 15µg/mL)Drug CI ORNet 1 layer, 2 neuronsTransfer functions TANSIG-PURELINAlgorithms Bayesian regularizationRegression equation Y = 1 X + 0.00094 Y = 1 X + 0.0012MSEPa 0.02257 0.02517RRMSEPa 1.70743 1.57856BCMSEPa 0.02451 0.02749Content courtesy of Springer Nature, terms of use apply. Rights reserved.Page 13 of 16Sakurand AlZakri BMC Chemistry (2024) 18:17 Detection limit (DL) andquantitation limit (QL).DL and QL were computed as described in ICH guide-lines, utilizing the slope of the calibration plot and the standard deviation of the response. Exceptional DL and QL values underscored the approaches’ sensitivity.AccuracAccuracy was assessed by computing the mean recov-ery % for three analyzed concentrations of pure CI and OR, employing their corresponding linear equations. Accuracy was further ensured in commercial formula-tions utilizing the technique of standard additions. The recovery values, inserted in Tables6, 8, demonstrated the approaches’ accuracy.PrecisionIntraday and inter-day precision were evaluated by com-puting the RSD% for three analyzed concentrations of pure CI and OR on a single day and over three days, respectively. RSD% values below 2 indicated excellent precision, as detailed in Table6.SpecificityDiverse mixtures with varying OR and CI ratios, as well as tablet samples, were subjected to analysis via the FSD and MC approaches. The recovery reported in Tables7, 8 substantiated the excellent specificity of the approaches for estimating CI and OR concentrations in both in-lab mixtures and tablet samples.Statistical evaluationStatistical comparisons were conducted among the pro-posed approaches, as well as against the reported HPLC method. T-test, F-test, and One-way ANOVA were employed for this purpose. The scores indicated no nota-ble differences between the proposed approaches and the reported HPLC approach, as summarized in Tables9, 10. The statistical analyses were performed utilizing the data analysis function in Excel 2019.Application ontablet andserum samplesThe established ANN, FSD, and MC approaches were applied to tablet formulations and serum samples to pre-dict the actual concentrations of CI and OR. The high linearity range of the elaborated approaches permitted the estimation of CI and OR in human serum, where Cmax = 2.6 µg/mL for CI [36] and Cmax = 10.5 µg/mL for OR [37]. Mean recovery percentages and RSDvalues were computed for all samples, as presented in Table8, thereby highlighting the exceptional performance of the proposed approaches in the estimation process.Table 6 The validation outcome and parameters of the established tools for CI and OR quantificationa The accuracy of 3 concentrations of CI (4μg/ml, 6μg/ml, and 12μg/ml) and OR (6μg/ml, 12μg/ml, and 20μg/ml)b Intermediate Precision and Repeatability are computed as the RSD% of 3 concentrations of CI (5μg/ml, 7μg/ml, and 13μg/ml) and OR (5μg/ml, 7μg/ml, and 15μg/ml)Tool FSD MCDrug substance CI OR CI ORLinearity range μg/ml 2–15 μg/ml 3–25 μg/ml 2–15 μg/ml 3–25 μg/mlWavelength nm 280.1 nm 314.2 nm 272.0 nm 306.2 nmLinear equation Y = 0.2732 X + 0.0031 Y = 0.0778 X + 0.002 Y = 0.2091 X + 0.0026 Y = 0.0571 X + 0.0042R2 0.9999 0.9998 1 0.9999Mean% ± SDa 100.63 ± 1.23 100.50 ± 1.02 100.20 ± 1.35 100.11 ± 0.83Repeatabilityb 0.94 0.91 1.10 1.41Intermediate Precisionb 1.10 1.75 1.02 1.45DL μg/ml 0.107 0.163 0.139 0.209QL μg/ml 0.324 0.495 0.422 0.634Table 7 Quantification outcome of CI and OR by the suggested tools in the produced mixesa Average of 3 repetitionsb A ratio that mimics the announced ratio in tablet formulationTool FSD MCDrugs concentration (μg/ml) CI: ORCI OR CI OR12:4a 98.92 100.19 98.81 100.9312:6 100.10 101.39 99.95 100.806:12 98.43 101.85 98.01 101.468:8b 100.20 100.08 99.17 99.62Average% ± SD 99.66 ± 1.25 100.88 ± 1.19 98.98 ± 1.08 100.70 ± 1.34Content courtesy of Springer Nature, terms of use apply. Rights reserved.Page 14 of 16Sakurand AlZakri BMC Chemistry (2024) 18:17Table 8 Quantification outcome of CI, and OR via the suggested tools in tablet formulation and enriched seruma Average μg/ml for the 3 repetitions of CI and ORb The standard addition is(4,5, 6μg/ml) for CIPH and ORNIc The computed value is the average % ± SD for 3 concentrations of the standard added with 3 repetitionsd Recovery% ± SD for the 3 repetitions of CI and ORTablet formulationTool ANN FSD MCDrug CI OR CI OR CI ORDosage/Tab μg/ml 500 500 500 500 500 500Assay μg/mla 500.75 500.15 501.25 500.20 496.55 496.95Recovery% ± SD 100.15 ± 1.67 100.03 ± 1.78 100.25 ± 1.04 100.04 ± 1.15 99.31 ± 1.66 99.39 ± 1.81Standard addition b.c 99.40 ± 1.88 99.74 ± 1.82 99.98 ± 1.59 99.85 ± 0.24 9.60 ± 0.78 99.65 ± 0.22Serum sampleTool ANN FSD MCDrugs ratio(μg/ml) CI: OR CI OR CI OR CI OR2.6:10.5d 98.52 ± 0.85 97.9 ± 1.19 95.37 ± 1.10 98.00 ± 0.33 96.41 ± 1.04 97.39 ± 0.3910:5 96.42 ± 0.69 95.82 ± 1.50 95.50 ± 0.69 97.40 ± 1.52 95.57 ± 0.68 96.73 ± 1.415:15 100.30 ± 1.04 100.26 ± 1.12 96.78 ± 1.04 98.13 ± 0.50 96.13 ± 1.50 97.90 ± 0.14Table 9 Statistical comparison between the elaborated UV tools and the reference HPLC tool for CI and OR estimation in tablet formsa C-18 column utilizing acetonitrile: water as the mobile phase with (45:55v/v) ratio, pH sets to 3.0 via O-phosphoric acid, flow rate 1ml/min, estimation at 299nmb n = 5c f(0.05)6.388, t(0.05)2.306Tablet formulation CI OR HPLCaANN FSD MC ANN FSD MC CI ORAverageb 100.16 100.57 100.07 100.86 100.53 99.82 100.45 100.28SD 1.82 0.86 1.59 1.70 1.08 1.49 0.94 1.45Variance 3.3282 0.7378 2.5243 2.8826 1.1607 2.2267 0.8884 2.1019t-valuec 0.3169 0.2034 0.4604 0.5862 0.3082 0.4935 – –f-valuec 3.7464 1.2041 2.8415 1.3715 1.8109 1.0594 – –Table 10 ANOVA (single factor) outcome for comparing the introduced procedures’ results and HPLC procedures’ results for estimating CI and OR in tablet formulationSource of variation Sum of squares DF Mean square F value P‑value F critCI Between gro 0.828673 3 0.276224 0.14774 0.929628 3.238872 Within gro 29.91456 16 1.86966 Total 30.74324 19OR Between gro 2.901019 3 0.967006 0.462027 0.712699 3.238872 Within gro 33.48746 16 2.092966 Total 36.38848 19Content courtesy of Springer Nature, terms of use apply. Rights reserved.Page 15 of 16Sakurand AlZakri BMC Chemistry (2024) 18:17 ConclusionIn this study, the Bayesian Regularization Network, Fou-rier Self-Deconvolution, and Mean-Centering transfor-mations have demonstrated their remarkable efficacy for the concurrent quantification of CI and OR in tablet formulations and serum samples. For the multivariate technique, the Bayesian regularization model was elected over Levenberg due to its superior performance and accurate prediction across various drug ratio mixtures. A Bayesian network with (1 layer, 2 neurons) configura-tion was chosen for the concurrent estimation of cipro-floxacin and ornidazole with excellent scores of MSEP, RRMSEP, BCMSEP, and mean recovery. By employing univariate techniques, FSD with FMHW:70 and MC in the range of (240.0–290.0) nm for CI and (260.0–340.0) nm for OR were both successful in resolving CI and OR overlapping while maintaining the signal-to-noise ratio and exhibiting exceptional sensitivity for the overlapped components. The confirmed sustainability of these meth-odologies through green certificate, AGP, and whiteness metrics underscores their suitability for routine analy-sis. Notably, these approaches eliminate the necessity for prior separation processes in drug determination, offer-ing simplicity, speed, and eco-friendliness.AbbreviationsANN Artificial neural networks.trainlm Levenberg–Marquardt algorithms.trainbr Bayesian Regularization.CI Ciprofloxacin.OR Ornidazole.FSD Fourier Self-Deconvolution.MC Mean Centering.AGP Assessment of Green ProfileAcknowledgementsNot applicable.Author contributionsAA Sakur: supervision, methodology, revision. DAZ Visualization, Validation Writing an original draft. All authors read and approved the final manuscript.FundingNo funding.Availability of data and materialsThe data used in this study are available from the corresponding author on rational request.DeclarationsEthics approval and consent to participateThis study was permitted through the Committee of Research Ethics in the Faculty of Pharmacy, Aleppo University, Aleppo, Syria. All participants signed knowledgeable informed consent assertions earlier than participating in this study. All described strategies were achieved according to relevant guidelines and regulations.Consent for publicationNot applicable.Competing interestsThe authors declare no competing interests.Received: 17 September 2023 Accepted: 12 January 2024Published: 23 January 2024References 1. Juliana NCA, Suiters MJM, Al-Nasiry S, Morré SA, Peters RP, Ambrosino E. The association between vagin*l microbiota dysbiosis, bacterial vagino-sis, and aerobic vaginitis, and adverse pregnancy outcomes of women living in Sub-Saharan Africa: a systematic review. Front Public Health. 2020. https:// doi. org/ 10. 3389/ fpubh. 2020. 567885. 2. Meis PJ, Goldenberg RL, Mercer B, Moawad A, Das A, McNellis D, et al. The preterm prediction study: significance of vagin*l infections. Am J Obstet Gynecol. 1995;173:1231–5. https:// doi. org/ 10. 1016/ 0002- 9378(95) 91360-2. 3. Bezhenar VF, et al. The effectiveness of a monopreparation containing a combination of ciprofloxacin–ornidazole in the treatment of infectious-inflammatory and dysbiotic vagin* diseases. Meditsinskiy Sovet. 2021. https:// doi. org/ 10. 21518/ 2079- 701X- 2021- 13- 207- 215. 4. Yashchuk AG, Naftulovich RA, Battalova GYu. Experience of using the drug orcepol WM in complex therapy in women with urinary incon-tinence. Effect Pharma. 2022;18(24):20–2. https:// doi. org/ 10. 33978/ 2307- 3586. 5. Indian pharmacopeia, volume 3, 6th edition, Government of India, New Delhi: the controller of publication. 2010; 1823–1824. 6. Marcus Y, Tal N, Ronen M, Carmieli R, Gurevitz M. Correction: the drug ornidazole inhibits photosynthesis in a different mechanism described for protozoa and anaerobic bacteria. Biochem J. 2017;474(24):4269. https:// doi. org/ 10. 1042/ BCJ20 160433_ COR. 7. British Pharmacopoeia, The stationery office on behalf of themedicines and healthcare products regulatory agency (MHRA)-Crown Copyright, 2009.-References-Scientific Research Publishing.” 8. Sweetman SC. Martindale: the complete drug reference. In: 37th ed. Pharmaceutical Press. 2011. 9. Krishna JR. Development and validation of RP-HPLC method for the simultaneous estimation of ciprofloxacin hydrochloride and ornida-zole in combined pharmaceutical dosage form. J Adv Pharm Edu Res. 2014;4(4):440–3. 10. CarolinNimila I, Balan P, Sathiya Sundar R, Ashok Kumar J, Rajasekar S. Simultaneous RP-HPLC estimation of ciprofloxacin hydrochloride and ornidazole in tablet dosage form. Asian J Res Chem. 2011;4(2):227–30. 11. Tunca AK, Karakaya D, Bulbul S. Developing and validation of impu-rity and the simultaneous quantity determination methods for tablet forms containing Ciprofloxacin HCl and Ornidazole. Pak J Pharm Sci. 2020;33(3):1105–14. 12. Damerakonda KS, Hima Bindu M, Swamy Damerakonda K, Hima Bindu M. A novel validated stability indicating simultaneous estimation of cipro-floxacin and ornidazole by reverse phase high pressure liquid chroma-tography. Int J Pharm Bio Sci. 2015;5(3):94–101. 13. Grewal AS, Patro SK, Kanungo SK, Bhardwaj SK. Simultaneous spectro-photometric estimation of ciprofloxacin and ornidazole in tablet dosage form. Int J Pharm Sci Res. 2012;3(8):2716–20. https:// doi. org/ 10. 13040/ IJPSR. 0975- 8232. 3(8). 2716- 20. 14. Krishna JR, Naga B, Sandhya SH, Prasad VV. Development and validation of UV spectrophotometric method for the simultaneous estimation of ciprofloxacin hydrochloride and ornidazole in combined pharmaceutical dosage form. J Adv Pharm Edu Res. 2014;4(4):405–8. 15. Sakur AA, Zakri DAL. Three new, UV spectrum filtration protocols for the synchronous quantification of ciprofloxacin HCl and ornidazole in the existence of ciprofloxacin-induced degradation compound. Heliyon. 2023;9(12):e22752. https:// doi. org/ 10. 1016/J. HELIY ON. 2023. E22752. 16. Gandhi SV, et al. Chemometrics-Assisted UV spectrophotometric method for determination of ciprofloxacin and ornidazole in pharmaceutical formulation. ARC J Pharm Sci. 2017;3(1):19–25. https:// doi. org/ 10. 20431/ 2455- 1538. 03010 05. 17. Rote AR, Saudagar RB. New analytical method development and validation of ciprofloxacin and ornidazole in human plasma by high Content courtesy of Springer Nature, terms of use apply. Rights reserved.Page 16 of 16Sakurand AlZakri BMC Chemistry (2024) 18:17performance thin layer chromatography. Pharm Methods. 2016;7(2):89–93. https:// doi. org/ 10. 5530/ phm. 2016.7. 13 18. Rote AR, Saudagar RB. High performance liquid chromatographic determination of ciprofloxacin hydrochloride and ornidazole in human plasma. Pharm Anal Chem Open. 2015. https:// doi. org/ 10. 4172/ 2471- 2698. 10001 03. 19. Wasylka JP. A new tool for the evaluation of the analytical procedure: green analytical procedure index. Talanta. 2018;181:204–9. https:// doi. org/ 10. 1016/j. talan ta. 2018. 01. 013. 20. Gałuszka A, Migaszewski Z, Namieśnik J. The 12 principles of green ana-lytical chemistry and the significance of green analytical practices. Trends Anal Chem. 2013;50:78–84. 21. Groß A, Lenders A, Schwenker AF, et al. Comparison of short-term electri-cal load forecasting methods for different building types. Energy Inform. 2021. https:// doi. org/ 10. 1186/ s42162- 021- 00172-6. 22. Nabipour M, Keshavarz P. Modeling surface tension of pure refriger-ants using feed-forward back-propagation neural networks. Int J Refrig. 2017;75:217–27. 23. Ebhota VC, Isabona J, Srivastava VM. Bayesian regularization in multi-layer perceptron artificial neural network model to predict signal power loss using measurement points. Int J Appl Eng Res. 2018;13(22):15836–42. 24. Yang R, Zhang Y. A method of low concentration methane measurement in tunable diode laser absorption spectroscopy and Levenberg-Mar-quardt algorithm. Optik. 2020. https:// doi. org/ 10. 1016/j. ijleo. 2020. 165657. 25. Kauppinen JK, Moffatt DJ, Mantsch HH, Cameron DG. Fourier self-decon-volution: a method for resolving intrinsically overlapped bands. Appl Spectrosc. 1981;35:271–6. 26. Rajni R, Usha G. Mean centering of ratio spectra as a new spectrophoto-metric method for the analysis of binary mixtures of Vanadium and lead in water samples and alloys. Res J Chem Sci. 2012;2(9):22–9. 27. Khanchi AR, et al. Simultaneous spectrophotometric determination of caffeine and theobromine in Iranian tea by artificial neural networks and its comparison with PLS. Food Chem. 2007;103:1062–8. 28. Lomillo MAA, Renedo OD, Mart’ınez MJA. Resolution of ternary mixtures of rifampicin, isoniazid and pyrazinamide by differential pulse polarogra-phy and partial least squares method. Analy Chim Acta. 2001;449:167–77. 29. Khayata W, Zakri DAL. Two simple spectrophotometric methods for the simultaneous determination of benzocaine and phenazone. Res J Pharm Tech. 2018;11(6):2507–11. https:// doi. org/ 10. 5958/ 0974- 360X. 2018. 00463.8. 30. Sakur AA, Zakri DAL. A new selective colorimetric method coupled with a high-resolution UV method for the consecutive quantification of three drugs in semi-solid preparations. Heliyon. 2022;8(10):11003. https:// doi. org/ 10. 1016/J. HELIY ON. 2022. E11003. 31. Khayata W, Zakri DAL. Two simple spectrophotometric methods for the simultaneous determination of amoxicillin trihydrates and flucloxacillin sodium. Res J Pharm Tech. 2017;10(5):1327–32. https:// doi. org/ 10. 5958/ 0974- 360X. 2017. 00235.9. 32. de la Guardia M, Armenta S, Cervera ML, Gallart-Mateu D. The importance of incorporating a waste detoxification step in analytical methodologies. Anal Methods. 2015;7(13):1–16. https:// doi. org/ 10. 1039/ C5AY0 1202C. 33. Kammoun AK, Khayat MT, Almalkia AJ, Youssef RM. Development of validated methods for the simultaneous quantification of Finasteride and Tadalafil in newly launched FDA-approved therapeutic combination: greenness assessment using AGP, analytical eco-scale, and GAPI tools. RSC Adv. 2023;13:11817–25. https:// doi. org/ 10. 1039/ D3RA0 1437A. 34. Nowak PM, Posłuszny RW, Pawliszyn J. “White analytical chemistry: an approach to reconcile the principles of green analytical chemistry and functionality. Trends Analyt Chem. 2021;38:116223. 35. International Conference on Harmonization (ICH), Q2B: validation of analytical procedures: methodology, 62, US FDA, Federal Register 1997. 36. Esposito S, Galante D, Barba D, D’Errico G, Mazzone A, Montanaro S. Cip-rofloxacin concentrations in human fluids and tissues following a single oral dose. Int J Clin Pharmacol Res. 1987;7(3):181–6. 37. Matheson I, Johannessen KH, Bjorkvoll B. Plasma levels after a single oral dose of 1.5 g ornidazole. Br J Vener Dis. 1977;53:236–9.Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims in pub-lished maps and institutional affiliations.Content courtesy of Springer Nature, terms of use apply. Rights reserved.1.2.3.4.5.6.Terms and Conditions Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”). Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. Byaccessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For thesepurposes, Springer Nature considers academic use (by researchers and students) to be non-commercial. These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personalsubscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence ora personal subscription(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used willapply. We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally withinResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will nototherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission asdetailed in the Privacy Policy. While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users maynot: use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent accesscontrol;use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or isotherwise unlawful;falsely or misleadingly imply or suggest endorsem*nt, approval , sponsorship, or association unless explicitly agreed to by Springer Nature inwriting;use bots or other automated methods to access the content or redirect messagesoverride any security feature or exclusionary protocol; orshare the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journalcontent. In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journalcontent cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or anyother, institutional repository. These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information orcontent on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Naturemay revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved. To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or impliedwith respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,including merchantability or fitness for any particular purpose. Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensedfrom third parties. If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner notexpressly permitted by these Terms, please contact Springer Nature at onlineservice@springernature.com mailto:onlineservice@springernature.com