Introduction. Information Technology (IT) is a promising and quickly developing scientific and practical field. One of its traits is its ability to provide big numbers of people with everyday access to information and tools. That is one of the reasons IT is a popular topic for discussion and research in modern medicine which is already being heavily investigated and applied in somatic medicine (e.g. diabetes, hypertension). There are also multiple projects that aim to develop tools and methods for helping people with psychiatric disorders, most common being affective disorders (bipolar disorder, depression)[Park et al., 2019]. However, meta-analyses have shown that many of such applications lack clinically validated evidence of their efficacy[Wang et al., 2018]. Here we wish to present our first attempt in search for one.

When planning our project and the accompanying studies, our thinking was that IT can not only help track measurements or activities performed by the patient but also provide information and help to a larger sample of people that consists of healthy individuals, those with preclinical symptoms and manifesting psychiatric conditions. So, our goal was to develop a tool (a mobile application for availability purposes) that can help measure the level of psychiatric symptoms in any given person and, without making any clinical diagnoses, offer that person information about probable conditions they might have and help they can get. Later we plan to add modes that will help the person using the application to do and track activities that can be helpful as additional methods for making their psychiatric condition better and more stable.

Since we wanted the application to be interesting and available for the majority of the population, we decided to make it a mobile game and to make tests inside of it less formal. The first test we included was the Lusсher test as it is a rather well-investigated in relation to mood disorders tool that can be used as a screening method [Barrick et al., 2002]. The first dimension of psychiatric conditions we examined was affective disorders since they are common and in many cases their severity does not prevent people from using applications such as ours.

Methods. The participants were 62 healthy respondents (age Me=22, 14 male and 48 female) that comprised a control group and 21 respondents with a diagnosis of an affective disorder (F3x, F4x, ICD-10) undergoing treatment at the Republican Research and Practical Center for Mental Health in Minsk, Belarus (age Me=44, 9 male and 12 female). For measuring the symptoms of depressed mood we used an QIDS-SR16 inventory with its results division criteria modified by Assanovich. The inventory consists of 16 items and helps divide the respondents into 6 groups based on the severity of symptoms: very low, low, relatively low, moderate, high and very high severity. QIDS-SR16 has the highest sensitivity at the "very low-low" end of the spectrum and we considered it appropriate for our goals since our ultimate aim was to detect disordered mood in the preclinical sample. We compared two groups based on the severity of depressive symptoms using Mann-Whitney U.

The short version of Luscher test includes questions about a color that is most or least preferred at the moment and choice of that color from 8 specific ones. Then, the application arranged the numbers, associated with colors (0-7: 0 - gray, 1 - dark-blue, 2 - blue-green, 3 - red-yellow, 4 - yellow-red, 5 - purple, 6 - brown, 7 - black) into an order of preference from most liked at the moment to least liked. We counted and compared the frequency of each color in each position using Kraskell-Woles criteria in two groups and compared it between individuals with varying levels of depressive symptoms as measured by QIDS-SR16 using χ-square.

The application we developed was on its face a "farm" gaming application in which user is asked to plant trees and vegetables, gather the foods and communicate with 2 characters. Such design allows us to add many other diagnostic tools in the future, such as tracking of different dialogue options when speaking to characters or tapping assessment. For now, however, such nuance were not assessed, but they performed a role of making the Lusсher test placed inside one of the dialogues with a character less visible. We added a "fake" question about colors that the user would prefer in order to also hide the test that was actually assessed. The respondents from control group performed the tests themselves and patients performed them with the help of a member of a research group.

All the results were assessed using SPSS v 20.0.

Results. For the first part of the study we compared patients to the control group based on their QIDS-SR16 scores. The scores differed significantly (Mann-Whitney U, p<0,001). Next, we measured the frequency of each color on each position in the order of preference in the control and patient groups. Then we compared these frequencies with the Kraskell-Woles criteria and it turned out the distribution of color frequencies on the 5th position differed significantly between groups (p<0,001). The graphic representation of frequencies is illustrated in pic.1.

Then we applied another approach that can better illustrate the situation in the general population and not in the strictly clinical setting: we mixed all the respondents and divided them into 6 groups based on the modified QIDS-SR16 criteria. Then we compared color frequencies and found that here the 4th and 6th positions were the most sensitive for the intergroup differences (Kraskell-Woles criteria, p<0,05). However, we needed to distinguish the color that was most prevalent in the positions 4 through 6 and was responsible for the differences found. To accomplish that we used the χ-square test. It turned out the frequency of the gray color was significantly higher in the 5th position in the patient group compared to controls (p<0,05); the proportion of the black color in the 6th position was significantly higher in the group with "very low" level of depressive symptoms compared to the "low" group; the proportion of the gray color in the 6th position was significantly higher in the "low" group compared to "very low"; and proportion of dark-blue color in the 4th position was significantly higher in the "low" group compared to "relatively low".

Discussion. The tendency of depressed people to choose darker colors as more preferred at the moment, rather unexpectedly, has not been previously well documented, compared to other phenomena tied to the Lusсher test[Novovic et al., 1993][Cohen, 1978][Garvey, Luxenberg, 1987]. However, we showed such tendency in clean groups of individuals with clinical diagnosis and without one, which makes our work on the mobile application scientifically based. It is apparent that having only the results from Lusсher test is not sufficient enough to statistically divide people using the application into groups and give them recommendations. However, we can add to it the results of other tests mentioned in the introduction section and combined, such tests can yield psychological and psychopathological traits of an individual that will in turn allow us to place the person using the application into one of the several groups and produce recommendations for them which will be based not on the diagnosis or its absence, which we cannot fully distinguish without clinical observation, but on the preclinical picture displayed by the person.

‍