Using machine learning-based analysis for behavioral differentiation between anxiety and depression
T. Richter, B. Fishbain, A. Markus, G. Richter-Levin & H. Okon-Singer
Abstract
Anxiety and depression are distinct—albeit overlapping—psychiatric diseases, currently diagnosed by self-reported-symptoms. This research presents a new diagnostic methodology, which tests rigorously for differences in cognitive biases among subclinical anxious and depressed individuals. 125 participants were divided into four groups based on the levels of their anxiety and depression symptoms. A comprehensive behavioral test battery detected and quantified various cognitive–emotional biases. Advanced machine-learning tools, developed for this study, analyzed these results. These tools detect unique patterns that characterize anxiety versus depression to predict group membership. The prediction model for differentiating between symptomatic participants (i.e., high symptoms of depression, anxiety, or both) compared to the non-symptomatic control group revealed a 71.44% prediction accuracy for the former (sensitivity) and 70.78% for the latter (specificity). 68.07% and 74.18% prediction accuracy was obtained for a two-group model with high depression/anxiety, respectively. The analysis also disclosed which specific behavioral measures contributed to the prediction, pointing to key cognitive mechanisms in anxiety versus depression. These results lay the ground for improved diagnostic instruments and more effective and focused individually-based treatment.
Code
Please find the Matlab package under this link. For installation unzip the file into your Matlab working directory. Once the file is unzipped, you should start with depressionAnxietyAnls.m file and feed the information required. When requested by the code, use one of the Excel data files. Dataset1.xslx contains sub-clinical patients’ data. Dataset2.xlsx consists of clinical patients’ data.
Please cite the following paper in any future publication using this package: T. Richter, B. Fishbain, A. Markus, G. Richter-Levin, H. Okon-Singer, “Using machine learning-based analysis for behavioral differentiation between anxiety and depression”, Scientific Reports, 10:16381, 2020.
Data
| DASS_group | classification to groups on the basis of the DASS results. 1=HA, 2=HD, 3=HAD, 4=LAD |
| ISC_Neutral | mean RT (ms) of switching sequences-mean RT of non-switch sequences (also named Internal Switching Cost), for the neutral version of the IST |
| ISC_Emotional | mean RT (ms) of switching sequences-mean RT of non-switch sequences, for the emotional version of the IST |
| ISC_EMO-NEU | ISC of the emotional version of the IST-ISC of the neutral version. |
| ISC_N-N | mean RT (ms) for neutral-neutral word sequence in the emotional version of the IST |
| ISC_E-N | mean RT (ms) for emotional-neutral word sequence in the emotional version of the IST |
| ISC_N-E | mean RT (ms) for neutral-emotional word sequence in the emotional version of the IST |
| ISC_E-E | mean RT (ms) for emotional-emotional word sequence in the emotional version of the IST |
| DP.50.Angry.Congruent | mean RT (ms) for EDPT condition of short presentation, angry facial expression, and appearance of dot in the congruent location to the emotional stimulus. |
| DP.50.Angry.Incongruent | mean RT (ms) for EDPT condition of short presentation, angry facial expression, and appearance of dot in the incongruent location to the emotional stimulus. |
| DP.50.Happy.Congruent | mean RT (ms) for EDPT condition of short presentation, happy facial expression, and appearance of dot in the congruent location to the emotional stimulus. |
| DP.50.Happy.Incongruent | mean RT (ms) for EDPT condition of short presentation, happy facial expression, and appearance of dot in the incongruent location to the emotional stimulus. |
| DP.50.Sad.Congruent | mean RT (ms) for EDPT condition of short presentation, sad facial expression, and appearance of dot in the congruent location to the emotional stimulus. |
| DP.50.Sad.Incongruent | mean RT (ms) for EDPT condition of short presentation, sad facial expression, and appearance of dot in the incongruent location to the emotional stimulus. |
| DP.1000.Angry.Congruent | mean RT (ms) for EDPT condition of long presentation, angrybfacial expression, and appearance of dot in the congruent location to the emotional stimulus. |
| DP.1000.Angry.Incongruent | mean RT (ms) for EDPT condition of long presentation, angry facial expression, and appearance of dot in the incongruent location to the emotional stimulus. |
| DP.1000.Happy.Congruent | mean RT (ms) for EDPT condition of long presentation, happy facial expression, and appearance of dot in the congruent location to the emotional stimulus. |
| DP.1000.Happy.Incongruent | mean RT (ms) for EDPT condition of long presentation, happy facial expression, and appearance of dot in the incongruent location to the emotional stimulus. |
| DP.1000.Sad.Congruent | mean RT (ms) for EDPT condition of long presentation, sad facial expression, and appearance of dot in the congruent location to the emotional stimulus. |
| DP.1000.Sad.Incongruent | mean RT (ms) for EDPT condition of long presentation, sad facial expression, and appearance of dot in the incongruent location to the emotional stimulus. |
| DPCongruency_happy50 | mean RT (ms) for incongruent trials- mean RT for congruent trials of EDPT condition of short presentation, happy facial expression. |
| DPCongruency_happy1000 | mean RT (ms) for incongruent trials- mean RT for congruent trials of EDPT condition of long presentation, happy facial expression. |
| DPCongruency_angry50 | mean RT (ms) for incongruent trials- mean RT for congruent trials of EDPT condition of short presentation, angry facial expression. |
| DPCongruency_angry1000 | mean RT (ms) for incongruent trials- mean RT for congruent trials of EDPT condition of long presentation, angry facial expression. |
| DPCongruency_sad50 | mean RT (ms) for incongruent trials- mean RT for congruent trials of EDPT condition of short presentation, sad facial expression. |
| DPCongruency_sad1000 | mean RT (ms) for incongruent trials- mean RT for congruent trials of EDPT condition of long presentation, sad facial expression. |
| MEMPriming_neg | the relative difference between the percentage of identification of primed negative words from unprimed negative words (also named priming effect), in the implicit memory test of the WIT |
| MEMPrimig_neut | the relative difference between the percentage of identification of primed neutral words from unprimed neutral words, in the implicit memory test of the WIT |
| MEMPriming_pos | the relative difference between the percentage of identification of primed positive words from unprimed positive words, in the implicit memory test of the WIT |
| MemoryImplicit_neg-neu | priming effect of negative words-priming effect of neutral words, in the implicit memory test of the WIT |
| MemoryImplicit_pos-neu | priming effect of positive words-priming effect of neutral words, in the implicit memory test of the WIT |
| ExpMemNegative | number of negative words explicitly remembered in the WIT |
| ExpMemPositive | number of positive words explicitly remembered in the WIT |
| ExpMemNeutral | explicitly remembered in the WIT |
| EXPMEM_Neg-Neu | number of negative words-number of neutral words explicitly remembered in the WIT |
| EXPMEM_Pos-Neu | number of positive words-number of neutral words explicitly remembered in the WIT |
| INTERcongruent.neg | mean RT (ms) for congruent trials with negative distractors in the FAFT |
| INTERcongruent.neutral | mean RT (ms) for congruent trials with neutral distractors in the FAFT |
| INTERcongruent.pos | mean RT (ms) for congruent trials with positive distractors in the FAFT |
| INTERincongruent.neg | mean RT (ms) for incongruent trials with negative distractors in the FAFT |
| INTERincongruent.neutral | mean RT (ms) for incongruent trials with negative neutral distractors in the FAFT |
| INTERincongruent.pos | mean RT (ms) for incongruent trials with positive distractors in the FAFT |
| InterferenceCong_Neg-Neu | mean RT (ms) for congruent trials with negative distractors-mean RT for congruent trials with neutral distractors in the FAFT |
| InterferenceCong_Pos-Neu | mean RT (ms) for congruent trials with positive distractors-mean RT for congruent trials with neutral distractors in the FAFT |
| InterferenceIncong_Pos-Neu | mean RT (ms) for incongruent trials with positive distractors-mean RT for incongruent trials with neutral distractors in the FAFT |
| InterferenceIncong_Neg-Neu | mean RT (ms) for incongruent trials with negative distractors-mean RT for incongruent trials with neutral distractors in the FAFT |
| Interpretation_NegSelection | percentage of selection in negative interpretations in the WSAP |
| InterpretationRT_Neg | mean RT (ms) for selection in negative interpretations in the WSAP |
| InterpretationRT_Pos | mean RT (ms) for selection in positive interpretations in the WSAP |
| Interpretation_Rtneg-pos | mean RT (ms) for selection in negative interpretations- mean RT (ms) for selection in positive interpretations in the WSAP |
| ExpectancySelection_neg | percentage of acceptance of negative situations in the FET |
| ExpectancySelection_pos | percentage of acceptance of positive situations in the FET |
| ExpectancyRT_negative.accept | mean RT (ms) for acceptance of negative situations in the FET |
| ExpectancyRT_negative.reject | mean RT (ms) for rejection of negative situations in the FET |
| ExpectancyRT_positive.accept | mean RT (ms) for acceptance of positive situations in the FET |
| ExpectancyRT_positive.reject | mean RT (ms) for rejection of positive situations in the FET |
| ExpectancyConfidance_negative.accept | mean confidence ratings for acceptance of negative situations in the FET |
| ExpectancyConfidance_negative.reject | mean confidence ratings for rejection of negative situations in the FET |
| ExpectancyConfidance_positive.accept | mean confidence ratings for acceptance of positive situations in the FET |
| ExpectancyConfidance_positive.reject | mean confidence ratings for rejection of positive situations in the FET |
| ExpectancyRejection | mean RT (ms) difference between rejection of negative situations and rejection of positive situations in the FET |
| ExpectancyAcceptance | mean RT (ms) difference between acceptance of negative situations and acceptance of positive situations in the FET |