No abstract

Background SymptomGuide Dementia (DGI Clinical Inc) is a publicly available online symptom tracking tool to support caregivers of persons living with dementia. The value of such data are enhanced when the specific dementia stage is identified. Objective We aimed to develop a supervised machine learning algorithm to classify dementia stages based on tracked symptoms. Methods We employed clinical data from 717 people from 3 sources: (1) a memory clinic; (2) long-term care; and (3) an open-label trial of donepezil in vascular and mixed dementia (VASPECT). Symptoms were captured with SymptomGuide Dementia. A clinician classified participants into 4 groups using either the Functional Assessment Staging Test or the Global Deterioration Scale as mild cognitive impairment, mild dementia, moderate dementia, or severe dementia. Individualized symptom profiles from the pooled data were used to train machine learning models to predict dementia severity. Models trained with 6 different machine learning algorithms were compared using nested cross-validation to identify the best performing model. Model performance was assessed using measures of balanced accuracy, precision, recall, Cohen κ, area under the receiver operating characteristic curve (AUROC), and area under the precision-recall curve (AUPRC). The best performing algorithm was used to train a model optimized for balanced accuracy. Results The study population was mostly female (424/717, 59.1%), older adults (mean 77.3 years, SD 10.6, range 40-100) with mild to moderate dementia (332/717, 46.3%). Age, duration of symptoms, 37 unique dementia symptoms, and 10 symptom-derived variables were used to distinguish dementia stages. A model trained with a support vector machine learning algorithm using a one-versus-rest approach showed the best performance. The correct dementia stage was identified with 83% balanced accuracy (Cohen κ=0.81, AUPRC 0.91, AUROC 0.96). The best performance was seen when classifying severe dementia (AUROC 0.99). Conclusions A supervised machine learning algorithm exhibited excellent performance in identifying dementia stages based on dementia symptoms reported in an online environment. This novel dementia staging algorithm can be used to describe dementia stage based on user-reported symptoms. This type of symptom recording offers real-world data that reflect important symptoms in people with dementia.

INTRODUCTION The Clinician's Interview‐Based Impression of Change Plus caregiver input (CIBIC‐Plus) has been widely used in dementia drug trials to evaluate cognition, behavior, and function. New trials of symptomatic drugs forecast renewed interest in this measure. METHODS To test its clinical meaningfulness, we examined how CIBIC‐Plus performed in two cholinesterase inhibitor trials compared to goal attainment scaling Scale (GAS) scores, a patient‐reported outcome measure. RESULTS Net goal attainment was seen for all but one GAS domains in subjects who improved on the CIBIC‐Plus. Subjects who improved initially on CIBIC‐Plus scores were likely to remain improved across all other outcomes for each trial's duration, except for Disability Assessment for Dementia scores. DISCUSSION The initial response to treatment, as assessed by CIBIC‐Plus, remained stable for most outcome measures. Even small CIBIC‐Plus improvement changes are associated with clinically meaningful change as assessed by GAS. Other tests detect decline better than improvement.

Background In people with dementia, neuropsychiatric symptoms (NPSs), especially agitation, are associated with worse quality of life and caregiver burden. As NPSs may vary with illness severity, knowledge of how people with dementia and their caregivers describe and rate the importance of agitation symptoms can improve the understanding of the clinical meaningfulness of the manifestations of agitation. The internet provides new opportunities to better understand patient experiences, as patients and caregivers increasingly look to Web-based platforms as a means of managing symptoms. Objective The aim of this study was to examine Web-based reports from a dementia symptom website to better understand the symptoms of agitation and explore how they are being targeted for monitoring by caregivers of people with dementia. Methods The Dementia Guide website hosts a Web-based database used by caregivers (97%) and people with dementia (3%). From its 61 dementia symptoms, users can select relevant symptoms that they deem important to monitor or track the effects of treatment. We employed a staging algorithm to determine if individuals had mild cognitive impairment (MCI) or mild, moderate, or severe dementia. Agitation was defined using terms consistent with the International Psychogeriatrics Association’s provisional consensus definition. We compared the proportion of people with NPSs and agitation across stages of dementia severity and studied how many agitation-defining descriptors were selected, and how often they occurred, by stage. Results As of March 2017, 4121 people had used the tracking tool, of whom 2577 provided sufficient data to allow disease severity staging. NPSs were tracked by 2127/2577 (82.54%) and agitation by 1898/2577 (73.65%). The proportion in whom agitation was tracked increased with increasing cognitive impairment: 68.5% (491/717) in people with MCI, and 72.50% (754/1040), 73.3% (378/516), and 90.5% (275/304) in mild, moderate, and severe dementia, respectively (χ 2 3 =54.9; P <.001). The number of NPS and agitation descriptors selected also increased with severity (median number of NPSs=1, 2, 2, and 3 for MCI, mild, moderate, and severe dementia, respectively, Kruskal-Wallis H Test H 3 =250.47; P <.001; median number of agitation descriptors=1, 2, 3, and 4, H 3 =146.11; P <.001). Conclusions NPSs and agitation are common targets for tracking over the course of dementia and appear more frequently with increasing disease severity. These common and distressing symptoms represent clinically meaningful targets in treating people with dementia.