Lung cancer is the number one cause of cancer deaths. Many early stage lung cancer patients have a resectable tumor, however, their cardiopulmonary function needs to be properly evaluated before they are deemed operative candidates. Pulmonary function is assessed via spirometry and diffusion capacity. If these are below a certain threshold, cardiopulmonary exercise testing (CPET) is recommended. CPET is expensive, labor intensive, and sometimes ineffective since the patient is unable to fully participate due to co-morbidities, such as limited mobility. In addition, CPET is done using a set of physical activities that may or may not be relevant to the patient's typical activities.This paper presents steps towards developing a solution to address this gap. Specifically, we present OOCOO, a mobile mask system designed to measure oxygen and carbon dioxide levels in respiration, as well as activity levels. Unlike state of practice, oxygen, carbon dioxide, and activity data can be continuously measured over a long period of time in the patient's environment of choice. The mask is capable of wireless data transfer to commodity smartphones. We have carried out initial work on development of an Android application to capture, analyze, and share the data with authorized entities.
People with developmental disabilities often face difficulties in coping with daily activities and many require constant support. One of the major health issues for people with developmental disabilities is personal hygiene. Many lack the ability, poor memory or lack of attention to carry out normal daily activities like brushing teeth and washing hands. Poor personal hygiene may result in increased susceptibility to infection and other health issues. To enable independent living and improve the quality of care for people with developmental abilities, this paper proposes a new wearable sensing framework to monitoring personal hygiene. Based on a smartwatch, this framework is designed as a pervasive monitoring and learning tool to provide detailed evaluation and feedback to the user on hand washing and tooth brushing. A preliminary study was conducted to assess the performance of the approach, and the results showed the reliability and robustness of the framework in quantifying and assessing hand washing and tooth brushing activities.
Early detection of children with autism spectrum disorder (ASD) has been of great interest to researchers due to an increase in the rate of autism incidence around the world. However, a diagnosis of ASD is still challenging to receive in a timely manner for the large-scale population because the current diagnostic practice requires considerable cost and time, and do not provide quantitative feedback. In this paper, we explore a new ASD screening method, namely Gaze-Wasserstein, that is non-invasive, fast, and widely accessible. Based on the gaze tracking and analysis, Gaze-Wasserstein is able to provide objective gaze pattern-based measurements for home-based ASD screening, and can eventually be deployed on any mobile technologies with a front camera. To test the performance of Gaze-Wasserstein, we conducted a pilot study with 32 child participants where 16 children have ASD and 16 children are typically developing. Evaluation results demonstrate the effectiveness and time-efficiency of our proposed method in the ASD screening, which indicate that our Gaze-Wasserstein is a promising autism screening approach in the clinical practice.
Depression is a serious health disorder. In this study, we investigate the feasibility of depression screening using sensor data collected from smartphones. We extract various behavioral features from smartphone sensing data and investigate the efficacy of various machine learning tools to predict clinical diagnoses and PHQ-9 scores (a quantitative tool for aiding depression screening in practice). A notable feature of our study is that we leverage a dataset that includes clinical ground truth. We find that behavioral data from smartphones can predict clinical depression with good accuracy. In addition, combining behavioral data and PHQ-9 scores can provide prediction accuracy significantly exceeding each in isolation, indicating that behavioral data captures relevant features that are not reflected by PHQ-9 scores. Finally, we develop multi-feature regression models for PHQ-9 scores that achieve significantly improved accuracy compared to direct regression models based on single features.
The patellar tendon enables fundamental insight regarding neurological health status. Clinically observed dysfunction may warrant escalation to more advanced and expensive medical diagnostics. Conventionally clinicians apply an ordinal scale to quantify reflex response characteristics. However the reliability of ordinal scales is a subject of debate, and even highly skilled clinicians have disputed the observation of an asymmetric reflex pair. An alternative is the use of the wireless quantified reflex system, which features an impact pendulum attached to a reflex hammer for providing precisely targeted levels of potential energy with a smartphone (iPhone) equipped with software to function as a wireless gyroscope platform that can email a trial sample as an email attachment by wireless connectivity to the Internet. With notable attributes of the gyroscope signal recordings of the reflex response of a hemiplegic patellar tendon reflex pair observed a feature set is developed for machine learning classification. Using the multilayer perceptron neural network considerable classification accuracy is attained. The research implications reveal the potential of integrating machine learning with a wireless reflex quantification system that applies a smartphone (iPhone) as a wireless gyroscope platform.
Consistent power and cost effective health monitoring has become the need of the hour especially for the unstable, chronically and critically ill. Here we present a novel architecture and algorithmic methodology combining the sensing subsystem, symptom summarization, and data transmission. Physiological parameters from multiple sensors feed into a severity quantizer and a subsequent multiplexer, the output of which is processed by the RASPRO engine to rapidly discover and alert any health criticalities. The architecture is optimized for communication and energy performance, and the algorithms result in lucid presentations to physicians. The whole system is the result of close collaboration between engineering and medical teams at one of the best known multi-disciplinary universities, building on a multi-terabyte more than a million patient Amrita Hospital Information System (HIS) database, and is being readied for deployment on a large telemedicine network of more than 60 nodes in the Indian subcontinent and parts of Africa.
Mobile symptom reporting apps can conveniently gather health-related information at low cost from day to day, fundamentally altering the relationship between patients, health data, and care providers. However, current mobile systems face a difficult trade-off between the quality of the information they collect and the burden placed on patients. In this paper, we propose an algorithm for adaptive system reporting designed for mobile platforms. This algorithm uses personalization, domain-specific knowledge, and Bayesian reasoning to reduce the number of questions required for accurate disability assessment, substantially decreasing demands placed on the patient. Following development of the algorithm, we validate it retrospectively using responses to the 12-item multiple sclerosis walking scale collected from 31 subjects with multiple sclerosis. Trade-offs between accuracy and response quantity are explored in detail. In this dataset, a 42% reduction in the median number of patient prompts was achieved without causing a single clinically relevant estimation error. A 75% reduction was associated with 4.45% clinically relevant estimation error. Given these promising results, future work will focus on prospective validation in multiple sclerosis and other clinical populations.
Blood Pressure (BP) is a crucial vital sign taken into consideration for the general assessment of patient's condition: patients with hypertension or hypotension are advised to record their BP routinely. Particularly, hypertension is emphasized by stress, diabetic neuropathy and coronary heart diseases and could lead to stroke. Therefore, routine and long-term monitoring can enable early detection of symptoms and prevent life-threatening events. The gold standard method for measuring BP is the use of a stethoscope and sphygmomanometer to detect systolic and diastolic pressures. However, only discrete measurements are taken. To enable pervasive and continuous monitoring of BP, recent methods have been proposed: pulse arrival time (PAT) or PAT difference (PATD) between different body parts are based on the combination of electrocardiogram (ECG) and photoplethysmography (PPG) sensors. Nevertheless, this technique could be quite obtrusive as in addition to at least two contacts/electrodes to measure the differential voltage across the left arm/leg/chest and the right arm/leg/chest, ECG measurements are easily corrupted by motion artefacts. Although such devices are small, wearable and relatively convenient to use, most devices are not designed for continuous BP measurements. This paper introduces a novel PPG-based pervasive sensing platform for continuous measurements of BP. Based on the principle of using PAT to estimate BP, two PPG sensors are used to measure the PATD between the earlobe and the wrist to measure BP. The device is compared with a gold standard PPG sensor and validation of the concept is conducted with a preliminary study involving 9 healthy subjects. Results show that the mean BP and PATD are correlated with a 0.3 factor. This preliminary study shows the feasibility of continuous monitoring of BP using a pair of PPG placed on the ear lobe and wrist with PATD measurements is possible.
Recently, there has been an increased use of wireless sensor networks and embedded systems in the medical sector. Healthcare providers are now attempting to use these devices to monitor patients in a more accurate and automated way. This would permit healthcare providers to have up-to-date patient information without physical interaction, allowing for more accurate diagnoses and better treatment. One group of patients that can greatly benefit from this kind of daily monitoring is asthma patients. Healthcare providers need daily information in order to understand the current risk factors for asthma patients and to provide appropriate advice. It is not only important to monitor patients' lung health, but also to monitor other physiological parameters, environmental factors, medication, and subjective feelings. We develop a smartphone, sensor rich, and cloud based asthma system called AsthmaGuide, in which a smartphone is used as a hub for collecting comprehensive information. The data, including data over time, is then displayed in a cloud web application for both patients and healthcare providers to view. AsthmaGuide also provides an advice and alarm infrastructure based on the collected data and parameters set by healthcare providers. With these components, AsthmaGuide provides a comprehensive ecosystem that allows patients to be involved in their own health and also allows doctors to provide more effective day-to-day care. Using real asthma patient wheezing sounds, we also develop two different types of classification approaches and show that one is 96% accurate, the second is 98.6% accurate and both outperform the state of art which is 87% accurate at automatically detecting wheezing. AsthmaGuide has both English and Korean language implementations.
Monitoring behavioral abnormality of individuals living independently in their own homes is a key issue for building sustainable healthcare models in smart environments. While most of the efforts have been directed towards building ambient and wearable sensors-assisted activity recognition based behavioral analysis models for remote health monitoring, energy analytics assisted behavioral abnormality prediction have rarely been investigated. In this paper, we propose a data analytic approach that helps detect energy usage anomalies corresponding to the behavioral abnormality of the residents. Our approach relies on detecting everyday appliances usage from smart meter and smart plug data traces in regular activity days and then learning the unique time segment group of each appliance's energy consumption. We focus on detecting behavioral anomalies over a set of energy source data points rather than pinpointing individual odd points. We employ hierarchical probabilistic model-based group anomaly detection [7] to interpret the anomalous behavior and therefore, detect potential tendency towards behavioral abnormality. We apply daily activity logs to evaluate our approach using two real-world energy datasets pertaining to staged functional behaviors, and show that it is possible to detect max. 97% of anomalous days with max. 87% of meaningful micro-behavioral abnormal events generating 1.1% of false alarms. However, we show that our detected abnormality can be meaningfully represented to different stakeholders such as caregivers and family members to understand the nature and severity of abnormal human behavior for sustaining better healthcare.
The goal of this research was to design an RFID based wearable platform capable of continuous heart rate monitoring for infants. Two TechnikTex P180+B conductive fabric electrodes with interconnects were integrated onto a baby onesie and connected to an RFID heart rate detection circuit that used an on-off keying modulation scheme to transmit heart rate data. The quality of the output signals obtained from the TechnikTex P180+B electrodes had 98.90% correlation with that obtained from a standard MediTrace-230 foam electrode. An antenna connected to an RFID reader captured the RFID tag information and fed the data into a Raspberry Pi processor that was programmed to compute the heart rate. The real time heart rate was calculated by using a four-heartbeat moving average window with an acceptable error rate of 3 bpm. The calculated heart rate was then communicated to a mobile app via Bluetooth. The app displayed the heart rate as a discrete value and also as a real time graph. A local alarm system integrated with the Raspberry Pi was used as an emergency backup system to alert in case the baby’s heart rate was not in the range of 80 bpm to 180 bpm and/or in case there was a communication failure between the RFID tag, RFID reader, processor or the mobile device.
Alcohol abuse causes 88,000 deaths annually. In this paper, we investigate a machine learning method to detect a drinker's Blood Alcohol Content (BAC) by classifying accelerometer and gyroscope sensor data gathered from their smartphone. Using data gathered from 34 intoxicated subjects, we generated time and frequency domain features such as sway area (gyroscope) and cadence (accelerometer), which were classified using supervised machine learning. Our work is the first to classify sway features such as sway area and sway volume, which are extracted from the smartphone's gyroscope in addition to accelerometer features. Other novel contributions explored include feature normalization to account for differences in walking styles and automatic outlier elimination to reduce the effect of accidental falls. We found that the J48 classifier was the most accurate, classifying user gait patterns into BAC ranges of [0.00-0.08), [0.08-0.15), [0.15- 0.25), [0.25+) with an accuracy of 89.45% (24.89% more accurate than using only accelerometer features as in prior work). Our classification model was used to build AlcoGait, an intelligent smartphone app that detects drinkers’ intoxication levels in real time.
Long-term rehabilitation opportunities are critical for millions of individuals with chronic upper limb motor deficits striving to improve their motor performance through self-managed rehabilitation programs. However, there is minimal professional support of rehabilitation across the lifespan. In this paper, we introduce an upper extremity rehabilitation system, the Quality of Movement Feedback-Oriented Measurement System (QM-FOrMS), by integrating cost-effective portable sensors and clinically verified motion quality analysis towards individuals with upper limb motor deficits. Specifically, QM-FOrMS is comprised of an eTextile pressure sensitive mat, named Smart Mat, a sensory can, named Smart Can, and a mobile device. A personalizable and adaptive upper limb rehabilitation program is developed, including both unilateral and bilateral functional activities which can be selected from a list or custom designed to further tailor the program to the individual. Quantitative evaluation of the motor performance from the QM-FOrMS is derived from fine-grained kinematic measurements. We ran a pilot study with three groups, including five baseline subjects (i.e., healthy young adults), six older adults and four individuals with movement impairment. The experimental results show that QM-FOrMS can provide the detailed feature during the unattended rehabilitation exercise, and proposed metrics can distinguish the evaluation results across group.
This paper describes the communication challenges associated with implementation of a multi-sensor artificial pancreas system on a smartphone device and presents solutions for potential communication losses. In particular, a standardized development framework on an Android device supporting multiple sensors is introduced in order to facilitate a multi-variable closed-loop AP algorithm. Package losses and communication link errors pose major challenges for the AP systems to overcome. The proposed solution offers a conservative model, wherein the controller application on the host smartphone requests status updates, resending of packages if necessary, and ultimately, warnings are issued to the user when data cannot be recovered.
Walking impairment resulted by various chronic diseases, disorders and injuries have been investigated using recent emerging wearable technology, for instance, gait assessment using inertial body sensors in 6-minute walk (6MW) for persons with Multiple Sclerosis (PwMS) to identify spatiotemporal features useful to assess MS progression. However, most studies to date have investigated the features extracted from movements of the lower limbs and do not provide a holistic gait assessment. A recent pilot study demonstrated that the holistic gait assessment such as evaluating the associations among lower and upper limbs provided better discrimination between healthy controls and PwMS. This paper is motivated by this and further aim to answer the following question: can we identify the temporal gait patterns in terms of the holistic gait assessment? Traditionally this suffers from the statistical property of the causality inference method adopted by previous study. We proposed a deep convolutional neural network (CNN) to learn the temporal and spectral associations among the time-series motion data captured by the inertial body sensors. A simulated model was developed to train the CNN, and then the trained CNN was adopted to assess the gait performance from a pilot dataset with 41 subjects (28 PwMS and 13 healthy controls). Experimental results are reported to illustrate the performance of the proposed approach.
The pandemic outbreaks including seasonal influenza demonstrate the continuous threat from multi-level pandemics and underline the need for robust preparedness and response. The goal of this paper is to develop a symptom surveillance system (S3H), providing supplementary data with public health data for long-term pandemic monitoring and prediction in large population. The system targets at high spatiotemporal monitoring of symptoms in public area including fever and cough distribution. Our preliminary study was conducted in the university library. Thermal imager and sound sensors based on smartphones were used to acquire the raw data. Image processing algorithms were used to calculate the temperature distribution in the certain population. For cough sound recognition, we take the Mel-frequency cepstral (MFC) as the feature of cough sound and kNN algorithm was performed for automatically recognizing the cough sound in a continuous recording.
Depression is the most common mental disorder and is negatively impactful to individuals and their social networks. Passive sensing of behavior via smartphones may help detect changes in depressive symptoms, which could be useful for tracking and understanding disorders. Here we look at a passive way to detect changes in depressive symptoms from data collected by users' smartphones. In particular, we take two modeling approaches to understand what features of physical activity, sleep, and user emotional wellbeing best predict changes in depressive symptoms. We find overlap in the features selected by our two modeling approaches, which implies the importance of certain features. Characteristics around sleep, such as change and irregularity of sleep duration, appear as meaningful predictors, as does personality. Our work corroborates prior results that sleep is strongly related to changes in depressive symptoms, but we show that even a very coarse measure has some predictive capability.
Working Memory Span (WMS) tasks are some of the most commonly used tools in cognitive psychology. These tasks involve the presentation of a set of items in a specified sequence. The participant is then asked to recall the sequence of items in forward or reverse order with performance quantified by the maximum number of items correctly recalled. Though initially the test was developed for verbal administration, now with modern computing, these tasks are generally administered via a digital interface. As a result, a variety of implementations have been developed but generally target either research or consumer domains. There has yet to be an offering capable of making the necessary transition from validation in the research lab, through implementation in the clinic, to distribution directly to the consumer. The main barrier has been lack of an open, approachable architecture, robust enough for basic scientists but also technically scalable for wide distribution to consumers. Recent advances in software technology provide the necessary infrastructure for the development of a robust offering. We have developed a WMS task, built on the latest in multi-platform gaming technology, to act as a blueprint for how future tasks can be developed, tested, and distributed. Our implementation describes some of the fundamental principles necessary for cultivation of a rich ecosystem of health diagnostic and therapeutic software.
Although there is growing interest in using mHealth approaches to provide just-in-time [JIT] intervention elements tailored to current emotional states in real time, critical evidentiary questions remain unanswered. This study was a 'proof of concept' evaluation to see if, in the context of a manualized stress management program, adding JIT intervention reminders via mobile technology would enhance outcomes more than the provision of random (untailored) reminders (both also relative to a measurement only active control condition). Individuals participated in a stress management course and carried a palmtop computer for 11 days (2 days acclimation and assessment, 1 week intervention, and 2 days post-treatment assessment). For the intervention week participants were assigned to one of three groups and either: (a) only completed self-report assessments multiple times each data (using ecological momentary assessment [EMA]; (b) completed EMA and received random reminders to use stress management skills; (c) completed EMA and received JIT tailored reminders to use stress management intervention elements when high stress or negative affect were reported. Individuals receiving JIT reminders reported stressful events less frequently, lower stress severity, less negative affect, lower levels of a stress biomarker (cortisol), less frequent eating, less alcohol consumption, less smoking and better sleep quality than the other two groups at post-treatment assessment. Tailored mHealth interventions designed in conjunction with ambulatory assessment protocols to provide JIT intervention elements can improve the efficacy of a standard stress intervention above and beyond any effects of simple reminders or over the intervention training. Implications for mHealth and JIT interventions are discussed.
Medication adherence is pivotal for effective health outcomes. One of the main reasons behind poor medication adherence is forgetfulness, and reminder systems are often used in addressing the problem. This paper presents MedRem, a novel medication reminder and tracking system on wearable wrist devices. The system is handy and interactive, and it is enriched with several useful features. To address the limitations of the tiny display size of the wrist devices, MedRem incorporates speech recognition and text-to-speech features along with clever interface design. Users interact with the system using voice commands as well as using the display available on the device. A dictionary based training approach is used on top of the state of the art speech recognition systems to reduce the errors in recognizing the commands from the users. The system is evaluated for both native and non-native English speakers. The error rates for recognizing voice commands are 6.43% and 20.9% for the native and the non-native speakers, respectively, when a off-the-shelf speech recognition system is used. MedRem reduces the error to nearly zero for both types of users through a dictionary based training approach. On average, only 1.25 and 15 training commands are required to achieve this performance for the native and the non-native speakers, respectively.
