Our vision for a smart health ecosystem comprises sensing, processing, and communications technologies (Figure 2). These technologies unobtrusively and continuously collect data about an individual’s health status and context, allowing for a long-term picture of health and well-being. The analysis of these data allows access to comprehensible data summaries, personalized behavioral support (eg, alarms, notifications, and advice), and connection to health care professionals and other sources of support (eg, family, friends, or caregivers) when required.

A smart health ecosystem may benefit people with chronic conditions and the health care system. Here we give an overview of these opportunities.

Matilda is 43 years old, lives with her 2 young children, and works full-time. She recently saw her general practitioner (GP) for low energy levels and stress. Her GP ran some tests and found that Matilda had an elevated fasting blood glucose level of 6.2 mmol/L and a BMI of 31 kg/m². In addition, Matilda reported that her mother had type 2 diabetes. Following tests, Matilda is diagnosed with type 2 diabetes and is told by her GP that she will need to take insulin as part of her management plan. She is also encouraged to make changes to her diet and increase her physical activity levels. Matilda is busy, leading her to skip meals and snack late at night once her children have gone to bed to ease her stress. She does not have time, or energy, to cook meals after her long days at work but wants to improve her health so she can be there for her children. Matilda liked the idea of personalized support and enrolled in the Smart Health Ecosystem.

Matilda was given access to a continuous glucose monitor (CGM), a smartphone app, a wrist-worn fitness tracker, and a set of Bluetooth weight scales. The CGM provides blood glucose readings every 5 minutes (288 per day), allowing Matilda to track her blood glucose levels and trends. Previous research has shown that self-monitoring of blood glucose is an integral component of effective treatment planning for many patients with diabetes taking insulin [41,42]. The fitness tracker collects data on her activity levels, heart rate, sleep time, and location of her daily activities. Matilda was happy that she could use her personal smartphone to access the app as she was not keen to carry another device around. When Matilda first signed into the app, she was asked questions to assess her personality profile, likes, and dislikes, and was prompted to set some goals; the subsequent choice of behavioral support was based on this information. Matilda feels motivated when she receives regular feedback, likes to eat Thai food, enjoys dancing, and has a competitive nature. She chose to set goals for weight loss, waist circumference, and physical activity levels. Using the app, Matilda can monitor her blood glucose levels to make informed decisions about food choices and physical activity in relation to her goals. She weighs herself once a week using the smart scales, and she can also input her waist circumference into the app. Progress toward her waist circumference and weight loss targets is congratulated with “points” for a competition in the app; Matilda tops the leaderboard most weeks and wants to stay there. The app also sends Matilda reminders of her progress throughout the week, motivating her to continue making positive lifestyle changes.

Over time, the system learns Matilda’s routine. Matilda usually goes food shopping on a Saturday afternoon. Using her location data, when Matilda is on her way to the supermarket, she gets a notification on her phone that prompts her to open her app for a personalized shopping list. The list contains options for quick-to-prepare recipes and ready-prepared products based on her dietary needs and weight loss target. When the system identifies that Matilda has some free time—based on her location and activity status, she receives prompts to try a web-based Zumba class—provided through the app. During the class, her heart rate is transmitted to the app; depending on her heart rate, she gets on-screen prompts to increase her effort. As her fitness increases, algorithms adjust her targets and activity difficulty.

Matilda permitted the app to send summary-level data to be sent to her GP. Her GP can review these data to see her progress and facilitate discussion at her next visit or initiate a referral to a dietitian for follow-up. Similarly, once referred to a dietitian, these data could be summarized and provided to the dietician to generate a more personalized care plan. The Smart Health Ecosystem allows Matilda to be an active participant in her own care and reduce the risk of health complications and additional clinic appointments.

Herbert is 71 years old, lives alone at home, and has heart failure. Over 1 week, Herbert’s symptoms worsened with increased shortness of breath, a cough, fatigue, and swollen feet; as a result, his granddaughter suggested that he go to the hospital. Herbert was admitted for 3 nights, during which time he was monitored closely and treated with diuretics to reduce his fluid retention. He felt much better after his treatment and was enrolled in the Smart Health Ecosystem prior to being discharged.

Once home, Herbert was encouraged to measure his weight daily using Bluetooth scales. If a weight measurement was not detected by midday, and if motion sensors detected that he was at home, Herbert would receive a notification on his smartphone reminding him to step on the scale. One day, the Smart Health Ecosystem detected that Herbert’s weight had increased by 3 kg and that he had not accessed his sensor-enabled medication box in the days prior; it is likely that he had not taken his medication and was consequently experiencing fluid retention. An activity tracker indicated that Herbert was sitting longer than usual, prompting a symptom questionnaire delivered and answered through a conversational agent (eg, Alexa). Herbert reported that he was again experiencing shortness of breath and was feeling tired. System algorithms were used to analyze these parameters in relation to Herbert’s standard measurements (collected by EMA multiple times a day) and prompted an intervention.

The conversational agent prompted Herbert to contact his GP within 24 hours and reminded him to take his medication. An alert with summary-level data was sent to his heart failure nurse specialist for review. Due to the event’s proximity to Herbert’s recent hospital admission, the nurse called Herbert via the system to check in and advised him to adjust his diuretic medication. Subsequent hospital readmission was avoided, resulting in reduced hospital costs and increased quality of life for Herbert.

Implementing a smart health ecosystem will only be possible when the approach is collectively accepted and adopted by consumers, clinicians, allied health care workers (eg, nurses, physiotherapists, and social workers), and other health care stakeholders [43]. People have varying perspectives on technology, often stemming from their past experience and the design and usability of devices [44]. User experiences are varied across, for example, different personalities, diagnoses, and physical capabilities; hence, features may be acceptable to 1 person and not another [45]. Likewise, features that are acceptable to an individual at one time, or place, may not be acceptable at another time or location.

Multimorbidity can significantly impact the design and effectiveness of smart health ecosystems due to the complexity of managing multiple health conditions. People with multimorbidity risk being bombarded with multiple alerts and overwhelmed by messages from their various health care professionals. A smart health ecosystem can support care coordination by facilitating seamless communication and information sharing among different health care professionals. LLMs can help tailor care plans for the individual’s specific needs and help schedule multiple appointments. However, participatory or co-design values users as “experts” of their own experience and can be used to ensure that a smart health ecosystem meets user needs and preferences [46]. To help meet user needs, agile science, an iterative and adaptive development approach, can be used to evaluate the incremental and iterative development of digital health programs through initial development, optimization, and open source “release” phases. Toward the development of adaptive, responsive, and individualized programs, agile science also involves cocreation with end users and health care professionals and using machine learning and artificial intelligence principles [47].

Passively collected data on internet of things devices are subject to inaccurate and inconsistent measurements and missing data [48]. Some technologies are still emerging and require stabilization, and the validity and accuracy of devices continue to be tested [49-51]. Further appraisal of the validity and accuracy of devices in people with chronic conditions is also required. Devices should be carefully selected; a framework for device selection that considers technical, user experience, health and digital health literacy, data, regulations, safety and privacy, and investment concerns is available to support this [52]. To provide access to safe and evidence-based digital health products (eg, wearables sensors and CGM), we need a strong regulatory environment for digital health products. This will also improve the commercial viability of new innovations. Smart health ecosystems will need to align with existing regulations and be nimble enough to pivot as the regulatory environment inevitably changes in the future.

The acceptability of a smart ecosystem and its constituent components is closely linked to security, privacy, and trust [53]. Such systems need to be robust and resilient for end users (eg, patients) and clinicians to feel confident in their everyday use. Technology infrastructure bears the risk of cyberattacks [54] and exploitation of inappropriately collected personal and sensitive data. This potential for breach is significant in the presence of networked sensors and digital devices, which make up a smart ecosystem. This is primarily due to the reliance on wireless communications (which makes eavesdropping more straightforward) and low-energy technologies (which are unable to implement complex security protocols on their own) [25]. Such risks, however, can be reduced by applying privacy-by-design principles, such as storing and processing data on edge devices only (without the need to send it over a communication network).

Data security issues must be fully addressed and comply with national and international standards or regulations such as the HIPAA (Health Insurance Portability and Accountability Act) in the United States and the General Data Protection Regulation, applicable to the European Union and European Economic Area. Further, the Medical Device Regulation is a regulation that sets out requirements for the safety and performance of medical devices, including certain digital health technologies. It defines rules for classifying and certifying medical devices, including software used for medical purposes.

To ensure data accountability, clear guidelines on storage, control, and ownership are essential. As a rule of thumb, only minimal personal data should be collected and stored. Data aggregates and the principle of k-anonymity can be used to mitigate risks of data breaches and help advance research by releasing such anonymous data sets to the public [55]. An issue of note here concerns how information is handled when it falls outside the scope of HIPAA regulations, for example, when a consumer uses third-party health apps or web portals to access health data for personal use [56]. These health apps are often hosted by external entities and may subsequently share user data with limited transparency and accountability. Importantly, most of these apps are not subject to HIPAA regulations and may share user activity data with other third parties involved in advertising and profiling [56]. A practical approach to adopt is a policy that clearly details a data host’s obligation to protect and secure stored data and their liability for any unauthorized access or data breaches. Similarly, individuals can take responsibility for who they give consent to use their data. It is crucial to recognize that sharing data across different jurisdictions and countries presents complexities as international hosts and suppliers may not adhere to the same regulatory frameworks [25]. Furthermore, blockchain technologies have the potential to tightly control and document data access; the ownership model and transparency of blockchain solutions allow patients to determine who can view data and any views and modifications are trackable on a public ledger [57].

In recent years, the cost of data storage has significantly decreased. Various software companies offer cheap cloud storage while guaranteeing high-speed data access and taking care of backups and security. However, centralized data storage can introduce bottlenecks and especially network latency when it comes to retrieving and storing data from consumer devices. A shift to decentralized storage methods, also called edge computing, can alleviate some of these issues as data is stored and processed locally without needing a network to communicate to a central server. In either case, data storage is now abundant and rich user data can be collected continuously. Parallel processing and ever more efficient algorithms allow us to make sense of this “big data” and use it to find patterns, monitor patient states, and inform diagnostics.

A smart health ecosystem should seamlessly integrate technologies into people’s everyday lives and routines; this is likely to lead to greater adoption and engagement. Theories such as normalization process theory [58], or social practice theory [59,60], provide lenses through which we can understand how interventions might build on, integrate with, or disrupt everyday life. Implementation approaches must prevent digital exclusion and inequities in health care access [61,62]. A key benefit of a smart health ecosystem is that it can provide more than just telemonitoring. Training on how to use the system may be required [45]; this training could be provided by the system itself by adapting to the users’ learning targets over time. Furthermore, by providing feedback on measurements and the consequences of actions taken by the user, the system can reinforce learning—promoting future independence. From the health care perspective, a stepped approach to rollout might allow for natural adaptation without disrupting standard practice.

From the health care perspective, the smart health ecosystem will need to work alongside existing patient management and electronic health care record systems. Moreover, we need to create care management approaches and infrastructure that allow practitioners to engage with the smart health system without increasing the burden. For example, a busy GP does not have time scheduled in their daily routine to follow up or track alerts that can be generated by devices worn or used by patients. The smart health system will need to seamlessly interact with patient management systems across the continuum of care to enable health care professionals to make informed, data-driven decisions without leaving their existing clinical workflows. Alternatively, we may have to restructure the way we deliver health care to adapt to the use of smart health systems. For example, nurse-led interventions are being used to support people to self-care at home [63]. Nurses or other allied health professionals may be well-suited to ensure case management via the smart health ecosystem. Many GPs also believe that business and financial pressures in general practice influence their choice of consultation mode and where their time is allocated for non–patient-facing activities [64]. Therefore, there is more work to be done around the best funding models to support the effective use and monitoring of smart health care as also a top priority for primary care and health policy. A defined reimbursement framework for smart health ecosystems will be needed to overcome barriers to commercialization and uptake by health care professionals and consumers. Moreover, appropriate reimbursement will be an important step to ensure equity of access, particularly for those most in need. A stepped approach to rollout might allow for natural adaptation without disrupting standard practice. The new era of digital health can shift administrative work to technology, but the transition needs to be carefully designed. Economic costs and evaluation

Health care resources are scarce; thus, many systems have focused on mechanisms that optimize the efficiencies and long-term sustainability of the health system. Understanding the economic costs and the potential gains for the patient, clinician, and health care system is vital. Smart health care may be cost-saving to individuals and the health system, but economic evaluations are yet to confirm this. We know that web-based models of care are cost-saving over time, ranging from immediate to over 9 years [65]. However, the ways in which a smart health ecosystem will generate cost savings may be more nuanced, and our understanding of these continues to evolve [66].

Future research is needed to understand whether a smart health ecosystem can reduce hospital presentations and admissions. Furthermore, we need future research to understand whether smart health care can reduce patient travel and parking costs and ever-growing waiting times to see health professionals, and sick days will help clarify the potential economic gains of a smart health ecosystem approach. To promote cost efficiency, we suggest using a modular approach whereby the user interface, sensors, and content can be added, removed, or modified depending on user needs and preferences.

The use of digital health systems, tools, and interventions is reshaping health care services; however, this is likely disproportionate to health needs. Evidence suggests that individuals residing in economically disadvantaged areas, vulnerable and remote communities with varied cultural and linguistic backgrounds, people with disabilities, and those with limited health and digital health literacy are less likely to access or benefit from digital health services [67]. On the contrary, vulnerable population groups stand to gain the most from these services. Natural user interfaces, such as voice assistants, lower the barriers of entry to health services for vulnerable populations, and the ability to scale health services using technologies, such as LLMs, increases their accessibility. While these population groups used to bear a disproportionate burden of health care needs and encounter additional barriers in accessing essential medical services, technologies can grant them both access to and provide services at lower costs. Acknowledging and addressing the digital divide through constant reconsideration is imperative in order to enhance equity in accessing telehealth services [67].