Table 3.

Self-contained device-based solutions for Alzheimer’s disease.

Sr. No.	Study	Participants	Intervention Used	Evaluation Metric	Measurements	Key Findings	Study Limitations	User-Centred Design
Diagnosis
1	Gonzalez et al., 2019 [45]	N = 23; Age: 18–60yrs	Smart Cupboard	Control group (young Adults) vs. Experimental group (elderly) Self-reported memory test and face-name pair test	Memory assessment Reaction time	Pearson correlation results showed a high similarity between the test results and experimental results. No correlation between memory and age was found.	The assumption that only one person uses the cupboard. No cognition problem observed among individuals Battery usage	No
2	Suzumera et al., 2018 [48]	N = 31 AD (Mean Age: 74.2 yrs) 15 MCI (Mean Age: 74.3 yrs)	Mobile application	MMSE and control group 48 HC (Mean Age: 74.6 yrs)	JustTouch application evaluated the rhythm tapping, response and reaction times of fingers on the screen.	Spearman’s correlation coefficient values show significant differences between the 3 groups in terms of contact duration, rhythm and response.	Only index finger movement evaluated. A sample size of MCI patients was the bare minimum needed for the statistical test analysis.	Yes
3	Lee et al., 2018 [49]	N = 14 MCI; Age: 20s and 70s	Play apparatus with smartphone	Control group in the 20s and 70s	Reaction time	The response rate of the elderly in their 70s was 300 ms slower than that of those in their 20s. Response time of the patients was 9.64 times slower as compared to healthy controls in their 20s while 5.53 times slower than those in their 70s.	The use of Bluetooth might hinder speed accuracy.	Yes
4	Bartoli et al., 2017 [50]	N = 20 AD; Age = 57–82 yrs HC = 20 Age: 53–90 yrs	Haptic feedback robot interface	MMSE and Rey’s Figure copy-REY-Clock Drawing test Healthy Controls	Visio-motor coordination, position tracking, stability control.	The experimental group was significantly slower than HC. The attentive metric was negatively correlated with reaction time.	Session time and dominating hand difference could have affected results. With practice, the mind learns so the second session of all participants was better.	Yes
Assistance
5	Fardoun et. al, 2015 [51]	N = 41, Age: 55–72 yrs	Face recognition of relatives using a smartwatch	Customized Stored data (pictures) on cloud	Tapping the watch and taking picture of the person standing in front of the patients	The system did recognize the people with moderate accuracy, but the patient had difficulty in reading the data and taking pictures.	Small smartwatch screen. Every time picture should be taken at a 90-degree angle, hardware limited (camera watch), reading difficulty, accuracy low, connectivity issues	No
6	Rudzicz et al., 2015 [52]	N = 10 AD; Age ≥ 55 yrs	Walking and interactive robot ED	Observed by caregivers and experimenter	Robot monitored ADLs of the patients through visual recognition	The robot tried to guide them if they missed a step and also tried to converse with the patients using NLP. However, the users did not interact with the robot often enough ignoring the robot prompts over 40% of the time.	Robot teleoperated, the caregiver had to intervene, AD patient could not understand complex guidelines, could guide only through simple tasks,	No
7	McGoldrick et al., 2019 [53]	N = 3 AD patients and their caregivers; Age: 59, 71, 74 10–12 weeks study duration	Mobile Application	Observation by caregivers and Tau-U acceptance and usability assessment	Reminders set by caregivers for the days and weeks.	One AD patient left the experiment after one week of the intervention being used. Others showed improvement in the memory performance	Limited people, assessment criteria mostly subjective, feeding of reminders should be done manually Difficulty turning off reminders	No
Therapy
8	Lyu et al., 2020 [55]	N = 18 AD and caregivers; 12 weeks duration	Smart robotic dog and wearable system	Caregiver Burden Inventory system and questionnaire assessment	Physiological parameters, mood evaluation, EEG signals,	The robotic dog could interact with the AD patients making their mood better, on stimulus provision music played. The system could detect pneumonia accurately. Caregiver burden reduced because of the system.	Low accuracy = 63%, qualitative assessment, many wearables, EEG ear sensor, smart watch motion sensors on legs pulse oximeter band, etc., cardiopulmonary on chest, internet required for communication.	Yes
9	Stoekle et al., 2016 [57]	N = 5; Age = 62–69yrs memory loss patients	Music Player	Observational readings (experimenter)	Direct observation of the patients using the music player	The device can independently be used, and the users were able to shortlist the songs of their liking	Dedicated hardware (iPad 4), a limited number of songs, limited sample people.	Yes

Table: Acronyms: AD: Alzheimer’s Disease; MCI: Mild Cognitive Decline; MMSE: Mini-Mental State Exam; HC: Healthy Controls.