**2.4 Critical process attributes (CPAs)**

To develop an optimal manufacturing procedure, all the critical process attributes including facilities, equipment, manufacturing variables, and material transfer

**Figure 1.**

*The fundamental approach of QbD for designing of a pharmaceutical dosage form.*

*Applications of Statistical Tools for Optimization and Development of Smart Drug Delivery… DOI: http://dx.doi.org/10.5772/intechopen.99632*

should be considered. Pulverization, homogenization time/mixing, type of mixer, and energy input are the major critical attributes in the manufacturing of novel dosage form. The process attributes using these associated factors require be identifying and carefully controlling to formulate batches with reproducible quality [17].

Size reduction of the material may be affected by the types of mill used. Different types of material need a special type of mill for pulverization such as lignocellulosic biomass material (like wood) required 'fine grinding' (less than 100 μm) [18] but in other studies, the 'fine grinding' has been used for particle sizes up to 1 mm [19–21]. The excess temperature during processing can increase the degradation of ingredients [22, 23], while less temperature can cause the failure of the process due to drug solubility issues [24]. Mixing speed and time is a critical attribute to develop a smart dosage form. For optimizing the mixing speed and time, the minimum needed time to dissolve the components and the maximum time of mixing can affect the viscosity of the product (causing product failure) and it should be identified [22, 25, 26].

#### **2.5 Critical material attributes (CMAs)**

The quantitative and qualitative information of active pharmaceutical ingredients (API) is prime attributes as material attributes [11]. Although an API is mostly incorporated at low concentrations and occupies a negligible part in the final formulation, the additives (inactive ingredients) usually elucidate the physical properties of a formulation [11, 27]. A number of researches have shown that additive(s) can influence the fate of an API in dosage form [28, 29]. Different grades of additives show a substantial effect on quality attributes of the final product as well as the API stability in the product [30]. Impurities in a raw material may show a detrimental impact on the stability of API/additives. Another prime challenge during the design and development of a novel dosage form is the compatibility of additives and API.

#### **2.6 Design of experiments (DOEs)**

The DOEs is not a replacement for experience, intelligence, or expertise; it is a precious element for choosing experiments systematically and efficiently to give dependable and coherent information [31]. DOEs are defined as "an organized, structured technique for deciding the relationship between attributes influencing a process and the output of the process" [32]. The DOEs can be applied for the screening of designs/experiments, comparative experiments, response surface methodology, and regression analysis [33].

#### *2.6.1 Common experimental designs*

In order to provide a logical relationship among the dependent variables and independent variables, experimental designs may be classified into four classes: a) screening designs, b) optimization designs [34–36] c) comparative experiments, and d) regression modelling.

#### *2.6.1.1 Screening of designs/experiments*

Screening of designs involves the selection of prime factors affecting a response. For the selection of experiments; fractionate factorial designs, the full factorial designs, and Placket-Burman designs are mostly used for screening because these designs have cost-effective advantages. These screening designs permit one to study various input factors with minimum numbers of experiments. However, these

designs also show some limitations that should be contemplated in order to impart a better interpretation of the effects of input elements on output responses [34–36]. Only the linear responses are supported by screening designs. Thus, if a nonlinear response is observed, or a more accurate phenomenon of the response surface is required or more complex design may be applied [37].

## *2.6.1.2 The factorial approach*

The full factorial and fractional factorial designs are generally used by the most of the researchers as an alternative methodological technique to standard relative randomized controlled trials (RCTs) and module designs, which has supremacy over both for determining the active elements of formulations. The factorial designs are employed to explore the prime impacts of critical factors and interactions among factors [38–42].

The common and simple full factorial design is the 22 factorial designs, where 22 is indicating two factors at two levels means the total run of experiments is four, which are located in 2-dimensional factor space at the rectangle's corners. If there are 23 factorial designs is applied then total eight experiments are mandatory which are located at the corners of an orthogonal hexahedron on a 3-dimensional space. If large numbers of factors are used at large numbers of levels then the number of runs needed to finish the task. To minimize the number of runs, the fractional factorial design should be used (i.e., ½ or ¼ of the real number of runs of full factorial design) [43–45].

**Table 2** shows the three factors at two coded levels 0 and 1, where 0 represents a low level and 1 represents a high level. In **Table 2**, the last column shows the response values of random variables. The main effect of any factor (A, B, C) or interaction (AB, AC, BC, ABC) is the difference of two means, the means of the responses corresponding to high levels and the means of responses corresponding to low levels.

When we compare the suggestions of **Table 2** with the suggestions of fractionate factorial design shown in **Table 3**. In **Table 3**, every-even numbered test experiment eliminated from **Table 2**. Again, factor effects are differences in mean responses. Even though, the prime effect for factor A is absolutely similar and opposite in sign from the interaction AB; i.e., A is aliased with -AB. Each result in **Table 3** is aliased with another result, having the prime result for B which is aliased with the evaluation of the overall average response. Therefore, every difference in means measures the difference of two results; e.g., A-AB. Had the half-fraction accompanied the odd-numbered test experiments been removed, every difference of means for a result would be evaluating the sum of two results; e.g., A + AB.


#### **Table 2.**

*Full factorial design with three factors at two levels.*

*Applications of Statistical Tools for Optimization and Development of Smart Drug Delivery… DOI: http://dx.doi.org/10.5772/intechopen.99632*


**Table 3.**

*Fractional factorial design of a full factorial design with three factors.*

Every fractional factorial design needs the aliasing of all or some of the factor effects. Many times the selection of fractional factorial designs is the unscientific that can lead to ambiguity, even wrong, conclusions about factor effects. Inversely, it is precisely the attentive selection of which fraction is applied that can increase the experimentation efficiently without the aliasing of main effects. **Table 4** shows another half-fraction of the full factorial design.
