Dr. Mohan Kale
Abstract:-
The purpose of the article is to highlight use of statistical techniques in Ayurvedic studies. It emphasizes the need of statistics in such studies and helps to decide the exact nature, type and time of a typical medical experimentation. In addition, the information gathered through such experiments can be analyzed by various statistical techniques. Further, the modeling of specific phenomena by the use of Monte Carlo and / or statistical methods can be useful for the complete and detailed study of a particular treatment. Due to the uncontrolled deforestation for various purposes, raw material for the preparation of Ayurvedic drugs is rarely available resulting in high cost of drugs and experimentation. In the last part, the article discusses the need to build virtual laboratories so that such type of studies can become cost effective in terms of money, time etc.
- Introduction
In recent years Ayurvedic treatments are getting their share of preference by many people over the other types of treatments. The main advantages of this treatment include the complete eradication of the disease without any side effects of the drug taken and unlike the cosmetic relief offered by other contemporary types of treatments, the Ayurvedic treatment give a permanent solution to your disease. However, many objections are raised against the Ayurvedic therapy (though some of them are found to be baseless) such as long duration of the treatment, subjectivity of the treatment i.e. the doses and drugs are specific varying from person to person, dietary restrictions and use of some poisonous substances in the preparation of Ayurvedic drugs. It is the consequence of these objections that the use of this therapy has become a limited one and people are reluctant to take this treatment and prefer the other types of treatments.
Hence, to nullify and answer these objections, one has to do systematic studies and careful analysis of result has become absolutely necessary. The statistical methods, being the scientifically proven methods, can be used to fulfill this need of systematic analysis.
The present article is divided into five parts: The first part gives the introduction while the need and choice of a specific type of experimentation in Ayurvedic studies has been discussed in the second part. The third part is devoted to consider how statistical tests are useful for analyzing the association between various factors on the effectiveness of the treatment. The method for testing the effectiveness of a newly developed drug/treatment is also indicated in this section. The projections and model building of various situations by statistical techniques has been discussed in the fourth part. The last part indicates the use of computer and statistics to increase the effectiveness of the studies. - Need and Scope of Statistics in Ayurvedic Studies
The question, which many times asked, is when to conduct experiments on animals (trials). The answer to this question is many folds, in general; it is better to conduct experiments on animals when the effectiveness of the treatment is under assessment. However, due to variety of practical difficulties, it is not possible to conduct the experiment every now and then and hence the above question can be reframed as when is it absolutely necessary to conduct the experiment.
When one has sufficient earlier knowledge and / or documented evidences about the effectiveness of the treatment giving clear nod in favor of the treatment, then one can skip experimentation. But when a claim is made about the better effectiveness of newly invented treatment for the cure of some disease, or new combinations of some well known materials are to be tried out, it becomes absolutely essential to conduct experiments on animals and check whether one really gets the same results as been claimed. Also according to various certifying agencies and bodies such as food and drugs association (FDA), USA, accept the claim for newly developed medicine only when it has been clinically proved. The clinical results obtained should be statistically significant. Hence it has become mandatory for each treatment to undergo a clinical trial and also if one wants to popularize his / her treatment, it is better to follow these types of protocols.
Once it is decided to conduct an experiment, a natural question which comes in is, how to experiment and analyze the generated information to draw useful conclusions. Here statistics play a very useful role in answering this query. It tells us how to experiment in a systematic way and how to analyze the generated information so as to draw conclusions.
If the units on which the experiments are to be conducted are homogeneous in nature, then the results obtained are clear and one can draw conclusions very easily. However, in general, it is difficult to get homogeneous unit in sufficient number and experimental units have variations among them. Now, if after the administration of a particular treatment to such non-homogeneous units, different responses will be obtained from them and one may not be able to conclude whether it is due to the effect of the given treatment or due to the variation present in these units. To draw point at home, let us take an example: Suppose a group of n individuals suffering from underweight problem is selected and enima (Basti) treatment is administered to all of them. Further, let n1 of them show weight gain and n2 show no change (n=n1+n2). Then it is difficult to conclude on the basis of this information only about the effectiveness of the treatment. This is because an individual has its own tendency to react and response to the treatment depending upon its bio-chemical make up. If all of these individuals are having the same bio-chemical make up, then surely n1>n2 implies effectiveness while n1<n2 tell the otherwise story. Hence, if units are not homogeneous, more scientific methodology is needed to analyze such situations.
Statistics fulfills this need with the technique called Design of Experiments demonstrating how to experiment and how to analyze the information. The technique is so powerful that one can make multiple comparison of the effectiveness of various treatments simultaneously. The technique works on its three basic principles
1. Randomization 2. Replications 3. Local control. Let us explain this in more details in case of Ayurvedic study.
In Ayurveda, individuals are divided into three main categories called Prakriti as Vata, Pitta and Kapha. Their combination is also three i.e. the individual can be Vata-Kapha, Vata-Pitta etc. However, for the simplicity of the discussion, only three types are considered here. Thus, there is first source of variation namely the type of the individual and depending upon the Prakriti, an individual may give a different response to the same treatment. If this variation is not there and if all of them are of the same type, then there could be other major basic factors, which influence the effectiveness, say Agni. All of these may not be same on this characteristic. Hence the first step is to single out various characters which are having significant influence on the effectiveness of the treatment. These will act as the source of variations. Now, depending upon the number of the source of variations, one may use any of the following designs:
a) Completely randomized design (CRD) : only variation due to treatment is the source variation.
b) Randomized block design (RBD): Two sources of variation, one due to treatment + one additional due to the non-homogeneity of the experimental units.
c) Latin Square design (LSD) : Three sources of variation, one due to treatment + two additional due to non-homogeneity
d) Balanced incomplete block design factorial experiments etc.
Consider the randomized block design with two sources of variations as indicated. The problem is to test whether all the treatments have the same effectiveness for the cure of underlying disease. Three principle of design can be applied as follows:
Suppose there are n experimental units / individuals having the following break up
Vaata Prakriti + Kapha Prakriti + Pitta Prakriti = Total
N1 N2 N3 N
Then all the individuals of the same Prakriti are grouped in to one group called block and each treatment is administered in each block. Which treatment is given to which individual is decided purely by random mechanism such as lottery methods by giving no preference to any one. Hence each treatment has replications equal to number of blocks. Then the data collected can be analyzed using F-test to conclude whether the effectiveness of the treatments is same or they differ. Similarly, working of the other designs can be discussed.
Further, it may be possible that there is interaction effect of the two treatments i.e. more natural and more realistic e.g. It is well known that Tulsi, Jesthamadh and Souph are recommended in the treatment for cough and cold. However what one observes is that the drugs, which are prescribed, are the combinations of these in aqueous state or they are in the form of chewable tables. This is because their interaction has more effectiveness than individuals as far as curing ability is concerned. If one is also interested in testing whether there is significant effect of interaction, then one has to work with what are known as Factorial Experiments. Another example is of a liquid drug called Bhallatakasava which is prepared from Bhallataka, Haritaki, Amalaki, Guduchi and Loha Bhasma taken in appropriate proportion. Now each one of them has its effectiveness for the deficiency or trouble, however, Asava prepared with these individual ingredients taking together has wonderful results and is handy to use than these individuals and is frequency used in Ayurveda. Similarly once can take the examples of Kumariasava, Drakshasava and others.
The examples discussed above are not new and everyone practicing in Ayurveda is already known their effectiveness. However, to draw the points, these examples are used so that the subject matter can be understood in a better way.
Another possible use of Statistics is to quantity the data generated. After a fair amount of period from starting a Ayurvedic treatment one wants to know whether the treatment is going in a right direction or not. The only way to know this is to undergo suitable Pathological tests, however, many times these tests give results which are qualitative in nature i.e. positive or negative ness of the test and not recorded quantitatively. For the analysis of effectiveness, it is absolutely necessary to quantify the data and condense and present it in proper form. This can be illustrated with a following example:
Consider a case of an individual under Ayurvedic treatment for the heart-related problem. After a sufficient time interval he/she undergoes a stress test and detected positive, then if the data are not quantitative, then one may think that the treatment is not useful. But if it is recorded that an individual has positive stress test for a comfortable walk of say 8 seconds before the treatment and after the treatment he/she has a stress test positive with a comfortable walking timing of 5 minutes, then one can immediately in which way the treatment is going. It may happen that after the second time, the test would be continued to be positive but with the comfortable walking timing of 8 minutes. Then this result sure indicates the effectiveness of the treatment. If this study is carried out for a group of individuals undergoing such treatment, then based on this data one can say something about the effectiveness of the treatment, in general. Same can be said about the lipid profile test and other tests for different diseases.
Lastly let us look at another problem, which has more chance of occurring in Ayurvedic studies called missing observation. This can be described as follows:
Suppose n individuals / units are selected for an experiment, which involves the study of the comparison of various treatments. The experiments are to be conducted for a specific duration. Now suppose due to negative canvassing or any other reasons, some individuals stop coming for the treatment and you have generated incomplete data with some missing observations. What one can do in such situation? Does all the efforts are in vain? Are we going to start a fresh? The answer is no. You do not have to do this once again. Statistics is at your help. The technique called the Missing Plot technique can help you to estimate these observations based on the data you have collected with a desirable accuracy. Using these estimated values at the place of missing observations, you can analyze the data and draw accurate conclusions. - Use of some Statistical Tests:
Many times in Ayurvedic studies one has to detect association between various factors to answer the questions, which are of the following types:
Does the effectiveness of the treatment depend upon the age?
Do the stage of the disease in which an individual is has any influence on the effectiveness of the drug?
Do I have to follow a strict vegetarian diet while undergoing an Ayurvedic treatment?
A typical set of complaints, are they indicator of the possibility of a specific type of disease uniquely? Or in other words, do I become susceptible to a particular type of a disease?
Do my typical type of life style make me more susceptible to a particular type of a disease? Etc.
To answer above types of questions, one has to undertake the study in which responses could be of the qualitative nature. The response of an individual could be any one of the following:
1. treatment is completely useful
2. treatment is partially useful
3. treatment is marginally useful
4. treatment is ineffective
now in this particular study, one has to analyze qualitative characteristics (attributes). To answer all above questions, it is recommended to use a technique for the analysis of two attributes, namely chi-square (2) test for the association of attributes. E.g. let us look at the question (a) Here the first attribute would be A: Age of an individual and the second, B: Response to the treatment. The attribute A can have four levels say Child, Young, Middle aged and old aged (defined by taking appropriate age group) also attribute has four levels labeled by appropriate names. Then, the data can be summarized in the form of a 4 (4 table called as 4 ( 4 contingency table, Then, one can use chi-square test of association of two attributes in 4 ( 4 contingency table to conclude about the association).
Similarly if one want to see whether there is any association between the dietary habits and effectiveness of the treatment, then attribute A has three levels: Vegetarian, non-vegetarian and mixed diet and the attribute B has four levels as earlier, then the data generated can be summarized in to 3 (4 contingency table and chi-square test for the association of attributes in 3 (4 contingency table can be use to analyze the data and conclusions can be drawn.
Another situation wherein has to make use of a statistical test is when the effectiveness of one single treatment is to be assessed. The experiment can be conducted as follows: A group of n individuals / objects suffering from a particular disease is selected and the status of their health before the treatment and after a sufficient time interval of a particular given treatment is noted generating the data in the variable form. Let the health status before and after the treatment be denoted by X and Y respectively, then the data generated would be (Xi Yi), I = 1,2, —- n wherein Xi = health status of ith individual before the treatment and Yi = Health status of ith individual after the treatment.
Evaluate di = Xi Yi = I = 1,2— n. Based on these di values one can use pair t-test to analyze the data and conclude about the effectiveness of the treatment.
Illustration: Suppose Basti treatment is recommended for gain in weight. A group of m individuals having underweight problem is selected. Let the weight before the treatment of each of these individuals be recorded as X1 —– Xn. Basti is administered to all of them and then after a sufficient time interval, let the weight after the treatment for each of these individual be Y1 — Yn. Then di = Xi Yi, I= 1,2 n, then using these values and pair t-tst, one can conclude whether Basti is effective for increasing the weight.
There could be a situation when one only wants to compare the effectiveness of two different treatments. There could be a situation in which it is required to test whether the Ayurvedic treatment is equally effective as compared to the other contemporary pathies for curing a typical type of a disorder. This is needed because many a times, false negative canvassing is done against Ayurveda by saying that our brand of treatment is better than Ayurveda and Ayurvedic treatment cant match our brand in power of curing the disorder. To answer this criticism, it is recommended to use what is called as two-sample t-test. The experimentation will be done as follows: Suppose we want to compare two brands, say A and B, a: Ayurvedic and B: Other brand. Administer treatment A to a group of n1 individuals and treatment B to a group of n2 individuals. N1 may or may not be equal to n2. Then using the data, calculate averages and dispersion for each of the two groups and using these and the mechanism of two-sample t-test one can conclude about the equality in effectiveness of the two treatments.
Similarly there are other Statistical tests, which can be used in Ayurvedic studies as and when required. - Projections and Modeling
It is many times thought that various substances used in Ayurveda to prepare Bhasma are poisonous and hence it is harmful to take such treatment. This has again turned out to be a baseless allegation against Ayurveda. However, one has disapprove this claim by using scientific methods. The typical examples are those of Mercury and Sulfur used in the preparation of Ayurvedic drugs curing various disorders. The experiment can be conducted as follows: Give different concentrations of such doses to the animals under observation and record their health status. It is well known that a specific type of enzyme secretion takes place in liver to encounter any poisonous element entering the body. Now, it is good to record the concentration of this enzyme secreted and its ability to nullify the poisonous effect. Now to know about the concentration of the above-mentioned drug that can be tolerated by the animal under study, is to keep on increasing the drug concentration till the animal dies and the dose at which an animal dies is the Threshold of tolerance. If this threshold is much higher than the amount of the element in the Ayurvedic drug, then the allegation is baseless. However, statistics provides a technique in which one can obtain this threshold without killing the animal. The procedure is as follows: Give various concentrations of doses to animals which are not too high and record the health status of the animal, denoted by X: Concentration of the dose and Y: Health status. Collect observations ( Xi, Yi), I = 1,2 — n on each of the n animals. Then using a technique called method of least squares, one can obtain model Y=f(x) i.e. Y is a function of x. Then one can substitute various values of X which are much higher and see what are the corresponding values of Y. Also the function can be plotted and using graphical procedure, one can obtain the threshold dose. Thus, it saves the animal (experimental unit) and allegation can be disproved as earlier by comparing two values. This exercise is nothing but modeling the phenomenon under consideration and projecting Y values for given values of X.
Second example wherein the projection and Modeling can be useful is that of deciding optimum dose of a treatment to cure a certain disorder. Let us explain this with an example: Again the illustration is taken to deliver a message rather than generating anything new. However, same experiment can be conducted for a new drug and conclusions can be drawn. Suppose an optimum dose for the treatment on diabetic disorder is to be decided. Further, let us assume that a new drug Madhumeharai is being designed to control this disorder and the optimum dose of this drug is to be found out. Now for various diabetic patients different intensity doses are being tried and data are recorded.
X= Intensity of dose to bring down the blood sugar to a controlled range (non-diabetic zone) and Y = Initial blood sugar content.
Then, using the method of least square technique, one can construct a model, X=g (Y), and for the given value of Y one can get the optimum value of the intensity of the dose.
Lastly, give a mechanism to answer the question, how long should I continue my treatment for specific type disorder? Here also one can use the technique discussed earlier to obtain optimum duration for the cure. The variable here are: X = Initial state of the patient and Y = Duration of the treatment required for the cure. - Use of virtual laboratory in Ayurvedic Experiments
With the arrival of an information technology generation, it has become more simpler and less time consuming to make experiments. What one can do is to conduct a sort of pilot experiment on handful of objects for a manageable time span. Then supply this information to computer in the form of suitable database and then keep on simulated observation give you the output as if you have conducted the experiments under given conditions with desired accuracy. In this exercise, lot of money, energy and time is saved and once can generate as many observations as one ants. Thee are not the true observations but the simulated ones. Hence the experiments involving simulations are called virtual experiments and the system is the virtual laboratory. This type of technique is very useful in conducting the trials on newly invented drugs. One can generate real data by using low intensity doses on animals and then by simulating, picture can be projected about the situations when higher dose is administered.
The discussion in earlier section about modeling and projection can be done very easily using computer and once a model is established it is easy to simulate the observation using this model. Hence, statistical techniques play a crucial role in constructing virtual laboratories.
In the Ayurvedic experimental studies also one can prepare a data bank of various Ayurvedic treatments which are available for various disorders and prepare a expert system to guide the treatment for a particular disorder and the effect of which can be tested using techniques mentioned above.
In all above discussion an attempt has been made to discuss the philosophy and logic behind various statistical techniques and their possible use in Ayurvedic experiments. The other details like formulae are avoided to make the discussion simpler and moreover, these can be obtained from various resources in the literature.
Last updated on March 9th, 2021 at 05:11 am