Detection of Mild Cognitive Impairment Through Natural Language and Touchscreen Typing Processing

Abstract

Mild cognitive impairment (MCI), an identified prodromal stage of Alzheimer's Disease (AD), often evades detection in the early stages of the condition, when existing diagnostic methods are employed in the clinical setting. From an alternative perspective, smartphone interaction behavioral data, unobtrusively acquired in a non-clinical setting, can assist the screening and monitoring of MCI and its symptoms' progression. In this vein, the diagnostic ability of digital biomarkers, drawn from Fine Motor Impairment (FMI)- and Spontaneous Written Speech (SWS)-related data analysis, are examined here. In particular, keystroke dynamics derived from touchscreen typing activities, using Convolutional Neural Networks, along with linguistic features of SWS through Natural Language Processing (NLP), were used to distinguish amongst MCI patients and healthy controls (HC). Analytically, three indices of FMI (rigidity, bradykinesia and alternate finger tapping) and nine NLP features, related with lexical richness, grammatical, syntactical complexity, and word deficits, formed the feature space. The proposed approach was tested on two demographically matched groups of 11 MCI patients and 12 HC, having undergone the same neuropsychological tests, producing 4,930 typing sessions and 78 short texts, within 6 months, for analysis. A cascaded-classifier scheme was realized under three different feature combinations and validated via a Leave-One-Subject-Out cross-validation scheme. The acquired results have shown: (a) keystroke features with a k-NN classifier achieved an Area Under Curve (AUC) of 0.78 [95% confidence interval (CI):0.68-0.88; specificity/sensitivity (SP/SE): 0.64/0.92], (b) NLP features with a Logistic regression classifier achieved an AUC of 0.76 (95% CI: 0.65-0.85; SP/SE: 0.80/0.71), and (c) an ensemble model with the fusion of keystroke and NLP features resulted in AUC of 0.75 (95% CI:0.63-0.86; SP/SE 0.90/0.60). The current findings indicate the potentiality of new digital biomarkers to capture early stages of cognitive decline, providing a highly specific remote screening tool in-the-wild.

Keywords: Alzheimer's disease; deep learning; fine motor impairment; keystroke dynamics; machine learning; natural language processing; remote screening; smartphone.

Figure 1

The dependency parsing of the sentence “I am the grandmother of a wonderful grandson” is depicted. The ids of the words, representing the position of the words in the sentence, are indicated with numbers 0–7 and the word “grandmother” is identified by the parser as the root word. The arrows have the direction from the head/root word to the dependent/children, the dependency distance between the pair is indicated on the arrow head and equals the difference of their id values. The words “the,” “am,” “I,” and “grandson” are dependent from the word “grandmother” and the words “wonderful,” “a,” and “of” are dependent from the word “grandson.” The MDD of this sentence based on the Equation (11) equals (3+2+1+4+3+2+1)/7= 2.28.

Figure 2

CNN model layout indicating input, convolution, flattening, dense, and dropout layers, output shape of each layer and number of parameters.

Figure 3

System pipeline and experiments' overview. Texts and typing sessions were collected from MCI patients and HC through the custom keyboard of the TypeOfMood application in a non-clinical setting. The total dataset was split accordingly (D1, D2, D3) to the needs of the three experiments that tested NLP models (EXP1), keystroke models (EXP2), and models with fused feature sets (EXP3). Multiple rounds of LOSO experiments yielded the best performing models to distinguish among MCI and HC.

Figure 4

Comparison of Receiver Operating Characteristics (ROC) curves of all the models in the first (A), second (B), and third (C–E) experiment. Area Under the Curve (AUC) values for models are shown with 95% Confidence Intervals. LR, Logistic Regression; RF, Random Forest; knn, k-Nearest Neighbors.