User:Ashishindani/sandbox

RADHIKa: Ratio-based analysis deriving basis for comparison of historical, parallel or interdependent reported ken of studies - a novel method for comparing interconnected and disconnected datasets is a method in multiple dataset analysis developed by Dr. Ashish Indani. This was initially published in 2016, October in the International Journal of Clinical Trials.

what is the method
The RADHIKa method is an analytical approach that can be used when data regarding both the effects (endpoints or outcome measures) and the Influencers (demographics or risk factors) are available. The philosophy behind this method is to analyze the effects first and then compare them by considering the factors that influence these effects. This type of methods is called "post hoc propensity-equated analysis ".

In order to apply the RADHIKa method, there are four major prerequisites:

Common effects factor: To compare multiple studies, the outcome measure parameter (effect) used to determine the safety or efficacy of the treatment must be common. The unit, mode of expression, and measurement technique should also be consistent. If different units of measurement or expressions are used, appropriate conversion should be applied. It is not mandatory for the outcome measure parameter to be a primary endpoint and can be a secondary endpoint in one or all of the studies.

Precursor relationship: RADHIKa calculations are based on assessing two major factors: the "effect" being compared to evaluate efficacy or safety, and the "influencers" that impact the effect. The relationship between a risk factor and its ability to influence the outcome or endpoint (the effect) is called a "precursor relationship." Careful identification and selection of the risk factors that influence the selected effect are necessary to establish the correct logic for comparison.

Parameter value and null expressions: All entities (effects and influencers) must be expressed as real numbers, including percentages and ratios. Amongst influencers, a maximum of two null or missing values is accepted. Null values are adjusted using the property of equal ratios, by adjusting the zero to the standard normal statistical expression. If there are more than two null values, the adjustments may lead to high deviation and vague results.

Defined objective and qualifications: Similar to any scientific experiment, objectives must be predefined before implementing RADHIKa analysis. The objective of analysis may have various dimensions, such as interval-based analysis of measurements or events, time-to-event analysis, effect time, and precursor analysis. The selection of the correct effect, its precursors, expression, sensitivity, and establishment of literature search criteria depends on the objective of the analysis.

How does RADHIKa method work
The RADHIKa methodology involves three main components:

Construction of Ken: Ken is a collection of data from predicate studies where the values of effects and influencers are tabulated uniformly. Based on the prerequisites for calculations, the predicates and variables to be analyzed are determined and used to construct the table. The required study data is obtained through literature search and then tabulated.

Calculations and Box Plot: Calculations involve determining the primary influencers ratio (R(In)) and mean influencers ratio (ψ). R(In) is calculated by dividing the value of an individual influencer in the predicate array by the value in the Part Under Evaluation (PUE) array. The mean of all R(In) values is the mean influencers ratio (ψ). The ratio of effects (R(E)) is calculated by dividing the value in the predicate array by the value in the PUE array. The range of equivalence is determined using the confidence interval boundaries of R(I)s. Finally, the absolute risk ratio or RADHIKa ratio (RR') is calculated, representing the ratio of R(E) with mean R(I) or ψ.

Plotting the box plots: Box plots are generated based on RR', the confidence interval boundary values for RR', and a z-value of RR. The box plot provides insights into the parameter's tendency and the difference between the two arms (PUE and predicate). The position of the box plot in relation to the line of unity indicates the performance of PUE, and the tails of the box indicate the significance of the outcome in each direction.

validity
The  RADHIKa   method  is validated by   performing calculations   for   randomized   studies and using the auto-control where the same dataset is used in calculation in the second arm.