Passive monitoring by smart toilets for precision health

Abstract

Smart toilets are a key tool for enabling precision health monitoring in the home, but such passive monitoring has ethical considerations.

PASSIVE CONTINUOUS MONITORING FOR PRECISION HEALTH

Precision health is an approach to prevent, diagnose, and monitor disease using information gleaned from an individual’s biological information (1, 2). Passive monitoring in a smart home setting, where appliances and devices are connected and controlled automatically, may provide such biological information. Whereas wearable health sensors must be actively applied and require a certain amount of human intervention, a sensor that performs passively and noninvasively can collect valuable health data in the background of everyday life. Of the possible locations in a smart home, the bathroom, and more specifically the toilet, is particularly well suited for such passive sensors. A “smart toilet” not only can monitor standard health characteristics, including temperature, heart rate, and oxygenation, through sensors in the toilet seat but also can collect biological samples (i.e., urine and stool) that contain useful health information. Measurements of volume and frequency can help to characterize functional disorders of voiding and defecation, whereas biochemical analysis of the excreta may identify diseases such as cancer at an early stage and enable longitudinal disease monitoring (3, 4). However, the reluctance to openly discuss excreta (5, 6) has hampered the development and acceptance of smart toilets, which hold the potential to seamlessly integrate urine and stool analyses as part of routine toileting events and to serve as a gateway to the “digitalization” of health care in the home.

CURRENT STATE OF THE ART OF SMART TOILETS

The use of human excreta to obtain data about markers to predict disease dates back to ancient India, when scholars noted that ants would rush to the urine of patients with suspected diabetes. Urine and stool tests are now routinely used for disease diagnosis and management but still require manual sample collection, which generally limits their use in the home. A smart toilet for health monitoring has been depicted in science fiction for decades, but only a few commercial products can automate collection and analyze health information from human excreta. Existing smart toilets are limited to systems that focus on improving the toileting experience, e.g., automatic lids, warmed seats, bidets, water-saving measures, and even wireless music streaming, rather than collection of health data. Recent development of remote health-monitoring systems (such as continuous glucose monitoring systems and wearable devices) has led to further technological advancements in the toilet space toward noninvasive means of delivering health care. Toilet seats can integrate sensors to measure nonexcretory data such as cardiac health (electrocardiogram), optical measurement of blood volume changes (photoplethysmogram), and measurement of ballistic forces exerted by the heart (ballistocardiogram) in addition to blood pressure and blood oxygenation (7). Furthermore, artificial intelligence and computer vision can be used to quantify the analysis of human excreta. For example, we have used optical scanners to automatically record excretory metrics such as the frequency of urination and defecation, urinary flow rate, and stool morphology, converting these previously subjective patient-recorded events into passively collected digital information (8). Similar optical analyses have also been developed by a research group at Duke University (9) and by industry, e.g., Toi Labs, OutSense, and Olive Diagnostics.

Recent proof-of-concept studies have additionally demonstrated automated sample collection within a toilet to facilitate molecular analysis. Tasoglu and colleagues have developed a toilet-mounted urine sample collector, which uses a hydraulic circuit with a flush water pump, sample pump, collector pump, sample collector, and collection vial turret to facilitate long-term routine use (10). Yellosis, a start-up backed by Samsung Electronics, is developing a urinalysis system that uses a mechanical arm to deliver a urinalysis strip toward the active urination site for simultaneous sample collection and processing (11). Coon and colleagues have reported comprehensive urinalysis using mass spectrometry (12). Grego and colleagues (9) have developed a “smart pipe” system for hands-free stool collection within the plumbing that separates solid from liquid and uses a spray erosion technique for stool sampling.

Because frequent laboratory measurements and clinic visits would pose a major burden on the patient and health care system, clinical studies on the utility of high-frequency health monitoring have previously not been feasible. New tools such as smart toilets may enable longitudinal studies that integrate high-frequency health monitoring to take place.

CURRENT BARRIERS AND CHALLENGES

There are myriad technological, behavioral, and ethical challenges in collecting data from the activity and products of human excretion (Fig. 1). Discussing toileting events, for example, remains socially unacceptable in many cultures. Patients can also be reluctant to handle their own excrement needed for existing home-based analysis. Home stool tests for colorectal cancer screening have been shown to reduce mortality; however, patient participation rates remain low, at 15 to 30% in the absence of active patient support efforts (13). A smart toilet could automate this process and negate the need for human handling of excreta. However, to harness the potential of molecular diagnostics using human excreta, efficient sample collection and preparation strategies must be developed (14, 15). Although some emerging technologies describe mountable devices that can be installed on existing toilets (7, 8), many systems require alterations to the toilet frame or the plumbing itself (10), which may increase the overall cost and hinder its widespread use. In addition, a toilet environment is inherently nonsterile, which may lead to false-positive or false-negative results in biochemical assays. Excreta sampling during active urination or defecation before the excreta mixes with wastewater or contacts the toilet bowl would prevent contamination but requires greater spatiotemporal accuracy. This process can be facilitated by computer vision to automatically recognize an excretory event and by mechanical arms to target sample collection to the site of active urination or defecation. Current stool collection methods use a plastic toilet insert to collect the stool before mixing with urine or wastewater, and such a platform can be integrated into a smart toilet. Indeed, many German toilets retain a “shelf” design that allows for visual inspection and potential sampling of stool before flushing. Antifouling surfaces, commonly used in microfluidics, can be implemented to induce self-cleaning and minimize contamination.

Fig. 1. The smart toilet as a key precision health tool.

Smart toilets and passive monitoring for precision health have the potential to make an impact on health care in the future. Smart toilets for the home are poised to offer quantitative analysis of urine and stool for precision health. Data from excreta collected by the smart toilet can be securely transmitted and stored in the cloud, where analyses for disease risk prediction and early intervention can be performed. There are still a number of engineering, social, and ethical challenges that need to be overcome before smart toilets for precision health become a reality. It is important to prioritize establishing an ethical framework for the use of smart toilets as their development for passive health monitoring continues.

Importantly, most smart toilet developments have focused on automated sample collection but not yet on automated sample processing. Whereas removing the need to manually handle one’s own excrement may lower one barrier, it is still cumbersome to package and ship the sample to an external laboratory for analysis. The introduction of countertop point-of-care assay devices may represent an intermediary step for home testing, where the user can take an automatically collected sample and insert it into a separate cartridge or device. Realization of the final stage in which sample collection and analysis are completely automated and hands-free may be challenging depending on the assay. Small-form point-of-care assays will be more feasible to implement and automate in a toilet environment. Urine dipstick tests and lateral flow assays, which are used in urine ovulation/pregnancy tests, use a paper substrate with reagent pads that react with urine analytes and commonly generate a colorimetric optical signal. Several such reagent pads could be multiplexed onto paper strips that are automatically dispensed within the toilet to interface with the urine stream. Microfluidic paper–based analytical devices have begun to emerge as a platform for small, low-cost urinalysis and could be used in smart toilets for the same purpose (16). Optical imaging devices could be used to quantitatively analyze a colorimetric or fluorescent signal, and the paper strip could then be discarded into the toilet. Automated stool testing will likely be more difficult to implement because stool typically must be homogenized before analysis, and assays requiring complex equipment and protocols (such as nucleic acid or microbiome analysis) will be difficult to implement within the form factor and constraints of a toilet environment (15).

In addition, the presumed advantages of early and accurate disease detection with continuous health monitoring compared with those with standard screening remain hypothetical. For example, blood in the urine (hematuria) can be associated with the presence of bladder cancer; however, the practice of mass screening and testing for asymptomatic microhematuria has not generated strong evidence for the improved detection of bladder cancer. More frequent monitoring may not always improve disease outcomes and crucially depends on the performance of the biomarker being monitored. The optimal balance between monitoring frequency, performance accuracy, and cost must be ascertained but, to do so, requires a platform that enables frequent monitoring of excreta (3, 14).

ETHICAL CONSIDERATIONS

Passive health monitoring and ambient intelligence present numerous ethical, legal, and social issues, particularly for privacy, data protection, fairness and representation, and consent (Fig. 1) (17). Sensors placed in a potential range of institutional or private settings, from hospitals to homes, raise overarching concerns regarding the collection of data from patients and third parties (e.g., health care staff or family members), as well as challenges for existing regulatory frameworks for privacy, consent, and safety. Smart toilets may heighten ethical challenges because of the greater social sensitivity and stigma surrounding excretion and toileting behaviors.

Smart toilets raise concerns regarding informational privacy. Prospective users retain rights to control data about themselves (e.g., images of stool and urinalysis results) and have personal privacy interests in limiting unwanted intrusion or observation (18). Personal privacy is especially salient in situations of vulnerability, including but not limited to using a toilet. Smart toilets also elicit concerns regarding public surveillance for health research or care visà-vis increasing use of sensors and passive monitoring in the home and other private settings. Even when images do not contain directly identifying information, people may consider images of their sensitive body regions, such as genital areas or anorectum, or toileting events inherently intrusive. Insofar as public disclosure of genitalia and toileting events violate social norms and etiquette in many cultures, prospective users could reasonably consider themselves vulnerable to exploitation if data generated from their smart toilet use are shared with those other than the research team or their treating physician. In formulating protections for smart toilet data and appropriate informed consent and transparency for users, it is critical for researchers and developers to keep these kinds of user sensitivities in mind. In addition, such sensitive data should always be encrypted during transmission. Cryptographic measures such as secure shell tunneling, public-key encryption when using native sockets, and encrypted emails should be mandated for communication. Whether smart toilets will be classified as consumer electronics or medical devices in the future, such data can be considered electronic protected health information and need to be created, stored, and transmitted in electronic format in compliance with the U.S. Health Insurance Portability and Accountability Act (HIPAA) of 1996.

Data protection and sharing are thus central issues to the design and deployment of passive monitoring devices. In the United States, HIPAA states that patient information is generally protected as “health data” when generated by health care providers (19). Given that commercial smart toilets placed in private or public settings may not fit the definition of health care providers, the data collected may not be protected as health data. At the same time, some jurisdictions may have relevant personal data privacy regulations, such as the European Union’s General Data Privacy Rule or California’s Consumer Privacy Act. Researchers and developers of health data sensors will need to consider applicable data regulations as well as provide required disclosures for patients and other users. For developers using biometric features, such as identifying users through their unique physical features (e.g., fingerprint or analprint) (8), they also need to be aware of applicable biometric data laws, such as the Biometric Information Privacy Act of the state of Illinois (20).

Local anonymization and data deidentification are key aspects of data protection, but there is an increasing risk of reidentification because of advancements in computing capabilities and availability of large public datasets (18). Efforts to protect privacy may also present trade-offs with the scientific value of the data. Researchers and developers of smart toilets, for example, may need to consider design measures that will more effectively preserve privacy, such as the type of imaging (e.g., thermal imaging that does not provide as much detail regarding body regions) or placing sensors to capture less identifying information, which would need to be balanced with the ability to collect data of sufficient scientific utility (17).

Informed consent for future research uses of data generated from smart toilets must address how the data will be stored, shared, and used. Storing or sharing of images involving sensitive body regions must be clearly conveyed to users. Researchers should also appropriately handle incidental findings from smart toilet analyses. As with other types of evolving health technologies, developers will need to address what types of findings will be reported to users and who can access data from such reports. Smart toilet users and research participants may also need to be informed whether additional findings will be reported as newer biomarker information becomes available. Empirical research and consultation with stakeholders will be useful for identifying appropriate strategies for consent and reporting of results.

Reports generated from smart toilet data will require careful consideration of appropriate access as well as how to enable transparency through notification for the range of potential users when third parties or business associates of covered entities under HIPAA access the data. Sensor data could potentially reveal criminal activities from the smart toilet’s urinalysis, for instance, illicit drug use. Communication and transparency regarding data collection are particularly important if protected research populations are among prospective users. Such issues also underline the need for researchers to heed applicable local reporting and data storage requirements (17).

The availability of traditional (“dumb”) toilets as an alternative to smart toilets has ethical implications from the point of view of ensuring that consent for smart toilet use is obtained. Consenting individuals at the point of smart toilet use is likely to reduce the durability of voluntary consent. This is particularly true in circumstances when a user’s need is urgent and no alternatives to a smart toilet exist. Thus, developers should consider these and other practical realities of smart toilet use that affect users’ free and informed consent and adapt consent models accordingly during the research design or pilot phases.

To enhance data quality, devices should be designed in ways that are physically or psychologically unobtrusive so as not to influence normal toileting behavior. It is understandable that developers would seek to avoid “alert fatigue” by minimizing the type and frequency of alerts. However, such alerts also serve to remind people that their toilet use is being monitored. Developers should thus implement designs that optimize data quality without sacrificing transparency (18).

People are generally more likely to accept some privacy intrusions when technologies generate social, scientific, or direct clinical benefit. Clear guidelines are needed to distinguish the potential benefits and harms of using digital biomarkers from human excreta. As with other types of passive monitoring, developers and researchers will also need to identify when there are potential users, such as children, who could present additional concerns regarding privacy and consent. Empirical research and stakeholder engagement are important for understanding which design features and data protection approaches reinforce users’ privacy preferences.

Toileting habits, gut microbiota, and other factors that could affect smart toilet data collection and analyses may be variable across different geographical areas, cultures, and subpopulations. To ensure that smart toilets provide benefit across different populations and communities, it is critical that researchers recruit and engage diverse participants, with particular attention paid to age, gender, underrepresented groups, and people with disabilities (17). For example, smart toilet developers will need to consider how data collection and analyses might be affected when a person who is menstruating uses the toilet. Since the U.S. Supreme Court decision in Dobbs v. Jackson [2022 WL 2276808; 2022 U.S. Lexis 3057] overturned the federal right to abortion in the United States last year, there have been examples of law enforcement or other government institutions gathering and searching people’s personal data to investigate and prosecute potential abortion cases. Smart toilet data could potentially be used for such purposes. Researchers and developers may need to inform users of these risks and consider offering options to restrict collection of pregnancy-relevant data. It is particularly important to think through how to make toilet technology usable to those whose disabilities alter how they fulfill their toileting needs. Cost presents another accessibility issue. Developers should use practices that promote design equity such that smart toilets are fit for purpose across populations with diverse toileting needs and cultural norms. For example, researchers could seek input from people with disabilities or older adults regarding sensor placement or desired types of analyses.

CONCLUSIONS

We believe that smart toilets will be a technology that enables accurate, reliable, and repeatable high-frequency human excreta measurements in a cost-effective and completely automated manner. In addition, smart toilets should facilitate the remote conduction of clinical trials and recruitment of a more representative patient population, an important benefit given that traditional clinical trials have a patient population biased toward those who are able to take time off work and repeatedly travel to clinical trial centers.

Despite the potential of smart toilets and passive monitoring for future precision health, their immediate health care impact will depend on technological advances and will hinge upon many unknown factors. Such factors include elucidating potential benefits over harms (e.g., privacy infringement), ensuring user acceptance and compliance, and investigating potential business models within an appropriate ethical framework. Among these, establishing the ethical framework for precision health should be at the forefront of the technological development of smart toilets for passive monitoring of health.