(185d) Data Pre-Treatment Analysis of Residence Time Distribution (RTD) Profiles for Pharmaceutical Manufacturing Applications

Given the numerous advantages of continuous manufacturing (CM) [1] in the pharmaceutical industry, there has been an immense research thrust aimed towards understanding process dynamics [2] for accurate process prediction and efficient implementation of CM. Amongst the various techniques [2, 3] implemented for this end, the characterization of residence time distribution (RTD) has been performed for various unit operations as well as entire manufacturing lines, to capture the overall system dynamics followed by a change in the processing conditions [4]. RTD profile of any unit operation in the pharmaceutical manufacturing lines, can be obtained by adding a pulse or step change of the selected tracer at the inlet of the unit operation and measuring the tracer concentration at the outlet as a function of time using Process Analytical Technology (PAT).

However, the tracer concentration profile can include significant intrinsic and extrinsic noises from various sources. The different sources of “extrinsic noise” include environment and process conditions. The sources of “intrinsic noise”, on the other hand, refer to the internal fluctuations of the system, which can be classified into the sensor, equipment, and material variability related to the inherent variability of the sensor during RTD measurement, equipment fluctuations during operation, and differences of tracer and bulk material properties, respectively [5, 6]. It is important to measure, quantify and remove the external noise as well as measurement variability from the RTD profiles obtained experimentally, as they can lead to inaccuracies in developing predictive models for evaluating system dynamics. However, the available literature focusing on the application of RTD for pharmaceutical manufacturing vaguely mentions enough details on this aspect of data pre-treatment of RTD profiles.

In this work, we aim to present and discuss various data pre-treatment and noise handling strategies aimed towards identifying effective strategies for denoising RTD profiles. The first aspect of the proposed work focuses on the quantification of the noise associated with the experimental RTD measurements. Following this, the second aspect focuses on handling the measurement variability and external noise for reliable RTD measurements. To this end, strategies similar to those available in image processing [7] and signal processing [8] are implemented. Some examples of noise filtering include linear and non-linear filters like adaptive filter, Kalman filter, recursive least-square, least mean square filter, and median filter. To reduce the variability, signal transforms, such as Fourier transform and Wavelet transform [9], are used for noise reduction, followed by low-pass or high-pass filtering. In the proposed work, the various data pre-treatment and noise handling strategies including time and frequency averaging, and exponential smoothing techniques, are evaluated and compared, to obtain a good signal-to-noise ratio (SNR) of the experimentally obtained RTD profile. Lastly, the effectiveness of different averaging and denoising techniques is analyzed to ensure that the important information pertaining to the RTD profile is retained. Thus, the proposed work aims towards developing guidelines for data pre-treatment strategies of RTD, which would be helpful towards effective utilization of RTD studies for predictive modeling of manufacturing lines.

References:

1. Schaber, S.D.e.a., Economic Analysis of Integrated Continuous and Batch Pharmaceutical Manufacturing: A Case Study. . Ind. Eng. Chem. Res. , 2011. 50: p. 10083–10092.
2. M. Ierapetritou, F.J.M., G. Reklaitis, Perspectives on the continuous manufacturing of powder‐based pharmaceutical processes. AIChE Journal, 2016. 62: p. 1846-1862.
3. V. Vanhoorne, C.V., Recent progress in continuous manufacturing of oral solid dosage forms. International Journal of Pharmaceutics, 2020. 579: p. 119-194.
4. Danckwerts, P.V., Continuous flow systems: Distribution of residence times. Chemical Engineering Science, 1953. 2: p. 1-13.
5. Kirsch, J.D. and J.K. Drennen, Near-infrared spectroscopy: applications in the analysis of tablets and solid pharmaceutical dosage forms. Applied Spectroscopy Reviews, 1995. 30(3): p. 139-174.
6. Blanco, M. and M. Alcalá, Content uniformity and tablet hardness testing of intact pharmaceutical tablets by near infrared spectroscopy: a contribution to process analytical technologies. Analytica chimica acta, 2006. 557(1-2): p. 353-359.
7. Brailean, J.C., et al., Noise reduction filters for dynamic image sequences: A review. Proceedings of the IEEE, 1995. 83(9): p. 1272-1292.
8. Vaseghi, S.V., Advanced digital signal processing and noise reduction. 2008: John Wiley & Sons.
9. Xu, Y., et al., Wavelet transform domain filters: a spatially selective noise filtration technique. IEEE transactions on image processing, 1994. 3(6): p. 747-758.