Statistical modeling of psychosis data

Abstract

Psychosis is a special type of mental disorder that affects around 2–3% of global population and has a strong genetic basis. Under psychosis, there is a group of diseases, which apparently may look alike and thus, it is difficult to isolate them from each other. Moreover, the credibility of real data related to psychosis is not only questionable due to its secondary nature but also its availability is grossly restricted because of the ethical constraints and prevailing social taboo. The present paper is a novel attempt to capture psychosis data by considering 24 input symptom constructs and 7 tentative responses (outputs) as per Brief Psychiatric Rating Scale-F2 (BPRS-F2). The captured input–output data as per Plackett–Burman design (PBD) of experiments (after consulting 40 psychiatrists) are statistically modeled, to determine their mutual relationships (i.e., outputs as the functions of inputs). Both Pareto-charts as well as normal probability plots are prepared to investigate the effect of each factor on different responses. Significant symptom construct(s) has/have been identified for each response. For example, emotional withdrawal has significant contribution towards schizophrenia, and so on. The psychosis data, thus collected, will be useful for further processing to extract more information of the said disease.

Introduction

Under psychiatry, psychosis is a complex type of mental disease, where patients frequently swing from reality to fantasy [1]. It has a significant prevalence rate in the global population [2] and a firm genetic basis [3]. Clinically, there are various types of psychotic disorders according to the aetiopatholgy and clinical features, such as schizophrenia, schizophreniform, schizoaffective, brief reactive, drug-induced, organic, mania, psychosis with depression, delusional disorder, psychosis not otherwise specified, etc. Some of the sign-symptoms are seen to be common in most of these illnesses and consequently, it becomes difficult to separately identify a particular illness from others [4]. As the sign-symptoms are generally imprecise in nature, clearly defined boundaries cannot be determined among various illnesses. Thus, there could be an overlapping of some illnesses [5]. Moreover, screening and diagnosis of psychotic illness are exclusively clinical and depend on doctor’s interpretations of the presenting sign symptoms, which may vary [6]. Moreover, many psychotic illnesses have frequent twists in their course with time and schizophrenia tops the list [7]. Moreover, one psychotic illness is frequently associated with other non-psychiatric illnesses, e.g., psychosis with depression [8], has variable aetiopathologies [3], [9], and thus there is no clear management plan. Such a complex picture of psychosis has made it an enticing interdisciplinary research area, although it has not received much attention till date.

The present paper is concerned about two hurdles related to psychosis—(i) poor availability of psychosis data and (ii) questionable data reliability. In society, psychiatric illnesses pose a social stigma [10], hence the diseases are reported much less compared to other diseases. This low reporting leads to scanty data available for the research. On top of this, stringent ethical measures in psychiatry and psychology [11] hinder the researchers to access data. As the patient is generally not in a position to give information about him, it is collected from his/her relatives. As a result, the validity of such data is highly questionable. Thus, valid data generation using statistical modeling technique could be a viable challenge not only to find out the relationships among the symptoms and corresponding disease outcomes, but also to generate valid data which can be further utilized in research. The present paper aims at it.

Data from experiments, testing of modules and products, simulation, epidemiological and market surveys, and quality control must be logically analyzed before drawing final conclusions [12]. Statistical modeling is considered to be the most significant technique, to do so [12]. The modeling technique consists of the following steps—(i) capturing the process variables, (ii) its statistical analysis, (iii) hypothesis testing, (iv) data interpretation, and (v) drawing reliable conclusions. As psychosis-related data are having questionable validity and poor accessibility [13], statistical modeling techniques could be suitable measures, where expert’s opinion can be incorporated into a proposed model to derive logical decisions regarding the relationships among the input process variables and the responses. Moreover, statistical modeling could be used to generate valid data from a pre-verified data model. Fortunately, present day’s statistical modeling tools and techniques have potential to handle almost all kinds of data collected through experiments, for reliability validation and data generation from a suitable model. Therefore, statistical modeling techniques are being actively used in different fields of research since last five decades and the path-finding contributions of Sir Ronald A. Fisher are legendary. He first introduced statistical analysis and design in agriculture, for reasonable data analysis, filtration of noise and model verification [14]. Factorial design and analysis of variance (ANOVA), two popular concepts of present day’s statistical design and analysis, were first introduced by him only. Afterward, statistical modeling techniques were not confined only in industrial research [15], rather it was unanimously accepted in various other research-fields like weather prediction [16], sociology [17], psychology [18], medical research [19] and many other disciplines involved in our day-to-day life. It is important to remember that unfortunately, till date, no modeling technique is reported to study psychosis.

In the present work, an attempt will be made to capture psychosis data considering 24 input symptoms and 7 responses (outputs) as per Brief Psychiatric Rating Scale-F2 (BPRS-F2). Input–output data will be collected after consulting forty psychiatrists according to Plackett–Burman design (PBD) of experiments and statistically modeled to establish their relationships. Moreover, significant input parameter(s) will be identified for each response.

The rest of the text is organized as follows. Section 2 details the method of psychosis data collection according to the PBD with the help of forty psychiatrists, Section 3 describes the principle of statistical regression modeling used in the present study to determine input–output relationships. Results are stated and discussed in Sections 4 Results, 5 Discussion, respectively. Some concluding remarks are made in Section 6.