Despite the global mental health crisis, access to screenings, professionals, and treatments remains high. In collaboration with licensed psychotherapists, we propose a C onversational AI T herapist with psychotherapeutic I nterventions (CaiTI), a platform that leverages large language models (LLMs) and smart devices to enable better mental health self-care. CaiTI can screen the day-to-day functioning using natural and psychotherapeutic conversations. CaiTI leverages reinforcement learning to provide personalized conversation flow. CaiTI can accurately understand and interpret user responses. When the user needs further attention during the conversation, CaiTI can provide conversational psychotherapeutic interventions, including cognitive behavioral therapy and motivational interviewing. Leveraging the datasets prepared by the licensed psychotherapists, we experiment and microbenchmark various LLMs’ performance in tasks along CaiTI’s conversation flow and discuss their strengths and weaknesses. With the psychotherapists, we implement CaiTI and conduct 14-day and 24-week studies. The study results, validated by therapists, demonstrate that CaiTI can converse with users naturally, accurately understand and interpret user responses, and provide psychotherapeutic interventions appropriately and effectively. We showcase the potential of CaiTI LLMs to assist the mental therapy diagnosis and treatment and improve day-to-day functioning screening and precautionary psychotherapeutic intervention systems.

We hypothesize that online movement videos have untapped potential for teaching physical skills, and we developed a platform that automatically generates practice plans from raw TikTok dance videos. The practice plans teach one segment at a time using fading guidance and part-learning principles and are presented using a web-based interface featuring concurrent visual aids. Two user studies (n=54, n=38) were conducted. The first showed significant improvements in learning outcomes compared to standard tutorials, underscoring the importance of well-structured practice plans and offering nuanced insights into the design and effectiveness of visual aids. The second study found that segmentation and emoji-based dual-coding only benefit learning when integrated into a well-designed lesson structure. We provide a set of practical recommendations for enhancing online movement learning, focusing on the need for substantive part-learning activities and careful use of visual aids to prevent cognitive overload.

The accurate estimation of peripheral capillary oxygen saturation (SpO2) is vital for monitoring respiratory health, with applications spanning medical diagnostics and fitness tracking. Remote photoplethysmography (rPPG) offers a convenient and non-contact approach for SpO2estimation. However, existing methods predominantly rely on data within the normal SpO2range, hindering their effectiveness during hypoxemia. Moreover, cross-user variations poses significant challenges for practicality. To address these limitations, we propose a simple yet effective normalization-based SpO2estimation algorithm. By aligning individual Ratio-of-Ratios (RoR) data with a standard model at the matching SpO2level, we mitigate cross-user variation, accommodate different camera configurations, and account for lighting changes. Our experiments demonstrate that the proposed method achieves an rMSE of 2.8% with leave-one-subject-out cross-validation across the full SpO2range (70%-100%), significantly outperforming existing RoR-based and CNN-based SpO2estimation approaches. Notably, our methods excel in accurately identifying hypoxemia, a critical clinical requirement. We anticipate broader applicability of our approach in rPPG-based vital sign monitoring, underlining the potential for enhancing robustness and reliability in various domains.

Accurate, continuous monitoring of human physical and physiological signals is critical to enhancing healthcare, personalizing education, and human interaction with the physical environment. Current methods for acquiring human data, however, frequently rely on cumbersome environmental instrumentation, extensive manual inputs, or uncomfortable rigid or adhesive wearable sensors. The emergence of computational fabrics has ushered in a new era of ubiquitous computing. By seamlessly integrating sensing capabilities into everyday clothing and accessories, we enable uninterrupted physical and physiological data acquisition and interpretation, providing a holistic view of an individual's status.

Kangaroo Mother Care (KMC), involving chest-to-chest skin contact between an infant and caregiver, is proven to be an effective intervention for preterm and full-term infants. Accurate monitoring of KMC duration and infant's vital signs during KMC is clinically important. Existing monitoring methods, however, rely on manual efforts and require rigid sensors or wires/electrodes on the infant's body. We propose Joey, a fabric-based approach to continuously monitor KMC duration and two vital signs essential to an infant's well-being: heart rate and respiration rate. Joey is a soft fabric necklace worn by the caregiver. It leverages the transmission of electrocardiogram (ECG) signals across individuals during skin-to-skin contact. With a minimalist fabric sensor structure, Joey measures KMC duration via the presence of mixed ECG signals. It then isolates the infant's ECG from this mixture with a proposed signal extraction algorithm and employs a diffusion-based denoising model to mitigate motion artifacts, enabling reliable inference of infant's vital signs. We fabricate Joey prototypes with off-the-shelf hardware and evaluate its performance with user studies. Results demonstrate that Joey achieves an average F1 score of 96% for KMC duration measurement, and clinically-acceptable accuracy in infant's vital sign estimation with a mean absolute error of 2.3 beats per minute and 2.9 breaths per minute in estimating heart rate and respiration rate. Clinical interviews further confirm the usability of Joey's sensing fabric for infant skin. A demonstration video of Joey is available at: mobilex.cs.columbia.edu/joey

Kangaroo Mother Care (KMC), involving chest-to-chest skin contact between an infant and caregiver, is proven to be an effective intervention for preterm and full-term infants. Accurate monitoring of KMC duration and infant's vital signs during KMC is clinically important. Existing monitoring methods, however, rely on manual efforts and require rigid sensors or wires/electrodes on the infant's body. We propose Joey, a fabric-based approach to continuously monitor KMC duration and two vital signs essential to an infant's well-being: heart rate and respiration rate. Joey is a soft fabric necklace worn by the caregiver. It leverages the transmission of electrocardiogram (ECG) signals across individuals during skin-to-skin contact. With a minimalist fabric sensor structure, Joey measures KMC duration via the presence of mixed ECG signals. It then isolates the infant's ECG from this mixture with a proposed signal extraction algorithm and employs a diffusion-based denoising model to mitigate motion artifacts, enabling reliable inference of the infant's vital signs. We demonstrate Joey's sensing capability with hand-shaking experiments, showing the real-time mixed ECGs. A demonstration video of Joey for actual KMC practice is available at: mobilex.cs.columbia.edu/joey

Conventional wisdom holds that laser-based systems cannot handle mobility due to the strong directionality of laser light. We challenge this belief by presenting Lasertag, a generic system framework that tightly integrates laser steering with optical tracking to maintain laser connectivity with high-velocity targets. Lasertag creates a constantly connected, laser-based tether between the Lasertag core unit and a remote target, irrespective of the target's movement. Key elements of Lasertag include (1) a novel optical design that superimposes the optical paths of a steerable laser beam and an image sensor, (2) a lightweight optical tracking mechanism for passive retroreflective markers, (3) an automated mapping method to translate scene points to laser steering commands, and (4) a predictive steering algorithm that overcomes limited image sensor frame rates and laser steering delays to quadruple the steering rate up to 151 Hz. We demonstrate Lasertag's tethering capabilty with various mobile targets, such as a VR headset worn during active game play, a remotely-controlled moving robot, and more. Lasertag paves the way for laser applications in highly mobile settings.

Locating underwater robots is fundamental for enabling important underwater applications. The current mainstream method requires a physical infrastructure with relays on the water surface, which is largely ad-hoc, introduces a significant logistical overhead, and entails limited scalability. Our work, Sunflower, presents the first demonstration of wireless, 3D localization across the air-water interface - eliminating the need for additional infrastructure on the water surface. Specifically, we propose a laser-based sensing system to enable aerial drones to directly locate underwater robots. The Sunflower system consists of a queen and a worker component on a drone and each tracked underwater robot, respectively. To achieve robust sensing, key system elements include (1) a pinhole-based sensing mechanism to address the sensing skew at air-water boundary and determine the incident angle on the worker, (2) a novel optical-fiber sensing ring to sense weak retroreflected light, (3) a laser-optimized backscatter communication design that exploits laser polarization to maximize retroreflected energy, and (4) the necessary models and algorithms for underwater sensing. Real-world experiments demonstrate that our Sunflower system achieves average localization error of 9.7 cm with ranges up to 3.8 m and is robust against ambient light interference and wave conditions.

We propose a new sensing technique for one-dimensional touch input workable on an interactive thread of less than 0.4 mm thick. Our technique locates up to two touches using impedance sensing with a spacing resolution unachievable by the existing methods. Our approach is also unique in that it locates a touch based on a mathematical model describing the change in thread impedance in relation to the touch locations. This allows the system to be easily calibrated by the user touching a known location(s) on the thread. The system can thus quickly adapt to various environmental settings and users. A system evaluation showed that our system could track the slide motion of a finger with an average error distance of 6.13 mm and 4.16 mm using one and five touches for calibration, respectively. The system could also distinguish between single touch and two concurrent touches with an accuracy of 99% and could track two concurrent touches with an average error distance of 8.55 mm. We demonstrate new interactions enabled by our sensing approach in several unique applications.

In this letter, the optimal and suboptimal nonlinear processing for correlation-based signal detection is addressed in symmetric alpha-stable noise. By maximizing the correlator output signal-to-noise ratio, a constrained functional optimization problem is established. As this optimization problem is hard to get analytical solution, we apply finite discretization to it and prove that the resulting approximation problem is a convex quadratic programming problem. This optimal nonlinear processing provides performance bound and design criteria for correlator detection. Based on the noise parameter α and the order statistic of received data, we further propose an adaptive method to determine the optimal threshold of the commonly used soft limiter. Simulation results show that the proposed method achieves near-optimal performance.

Energy efficiency has now become a key pillar in the design of communication networks. With millions more base stations and billions of connected devices, the demand for energy-efficient system design and operation will be even more compelling in the fifth generation (5G) communication network.In this paper, we focus on maximizing the downlink system energy efficiency (EE) of the 5G ultra dense network through efficient resource allocation algorithms. First, a constrained EE maximize problem is formulated. However, the resulting optimization problem is challenging due to its non-convex nature. To overcome this issue, the transformations based on fractional programming theory are applied, so that the primal problem can be converted into a convex form. Then, we adopt a centralized algorithm to solve the optimization problem and get the global optimal EE. In order to reduce the complexity, an efficient distributed algorithm based on alternating direction method of multipliers (ADMM) is further proposed. The simulation results show that both the centralized algorithm and the distributed algorithm converge to the same EE, but the latter one has lower computational complexity.

Impulsive noise modeled as symmetric α-stable SαS distribution is commonly seen in many practical communication scenarios. In this paper, we focus on the synchronization of single-carrier signals in the mixture of SαS impulsive noise and Gaussian noise. We first derive the Cramér-Rao lower bound (CRLB) of the joint estimation of carrier phase and timing offsets for both liner modulation (LM) signals and continuous phase modulation (CPM) signals. In order to minimize the CRLB, the optimal training sequence (TS) is then designed for the synchronization of the LM signals and the CPM signals, respectively. We further propose a practical low-order cross-correlation based synchronization algorithm, which can be used for both LM and CPM signals. Numerical simulations show that our designed TS can outperform the pseudo-random (PN) sequence. By taking minimum shift keying (MSK) signals for instance, simulation results show that the performance of our proposed synchronization algorithm has negligible gap with the CRLB.