Ideally, the patient’s health should be seen as the main objective of the health care network. Nonetheless, each actor may follow their own personal objective while trying to gain power and fearing losing it while providing or receiving health care.

We assume that the patients’ primary goals are to maintain or improve their health, manage their lives, and receive adequate support after receiving a diagnosis. A hierarchical or linear organization of health care (ie, putting the more medically educated HCPs at the beginning or top and the patient at the receiving end) might be outdated. Recently, patient-centered care has been considered to be more beneficial [45]. Given the trend of patient empowerment, it is natural to picture the empowered patient in the center with supporting individuals and HCPs acting in the surroundings. Regarding shared decision-making, most patients prefer to be included in the decision-making process.

Nonetheless, approximately half of all patients would prefer that the physician should have the final say, even though this case is legally unenforceable since physicians are merely acting on behalf of the patient. Therefore, the average patient strives to be included in the decision-making process, yet the preferences regarding the final decision would theoretically vary [46]. Furthermore, the patients’ preference for playing an active role in the treatment process depends on different factors, including their age [46], ethnicity, or level of education [47]. Some patients do not want to be confronted by their health condition or do not feel that they are sufficiently ill to take action, leading to a potential delay in reaching out to HCPs [48-50]. Hence, patients may differ in their preferences concerning decision-making or being educated. Nevertheless, the patient, or their legal guardian, is obligated to decide, which emphasizes the importance of sufficient education and information.

In Figure 3, we also included artificial intelligence (AI) as a relevant interacting node. Recently gaining popularity, AI is already part of various health-related apps and devices [51]. AI may also be integrated into mHealth, yet current algorithms are still too prone to errors. Hence, at this point in development, they cannot be trusted with medical decisions (eg, due to not being able to interpret social cues or engaging in personal discussions with the respective patient) [52]. Nonetheless, the continuous improvement of such apps justifies the use of AI in an assisting manner. In this context, the transparency and explainability of these systems is crucial to ensure human users’ trust in AI as assistants [52,53]. An explainable app assists the users and reveals why it made a specific assessment. A user with domain knowledge can then interfere and perform error correction in case of erroneous results. This approach may also help to detect uncommon lesions, which might be overseen due to the inexperience of the individual clinician.

Moreover, the AI screening approach may save time for the clinician, which can be redirected to patient consultation and more direct contact. Using a conversational interface, chatbots can, for example, provide an initial evaluation of a patient’s mental or physical state or empower individuals confronted with a life-altering condition [54,55]. Besides alleviating the frequent concern of patients to lose personal contact due to digital medical approaches [56], such approaches linking the AI to the clinician can further boost the doctor-patient relationship. Moreover, while the user benefits from an AI assistant without giving up control, the AI benefits from the chance of improvement through error correction [53]. In this sense, this interaction can be seen as mutually empowering if accessibility is put into practice to counterbalance social inequalities. Yet, it is essential to note that additional challenges accompany the benefits of AI integration. For this purpose, the WHO guidance on Ethics & Governance of Artificial Intelligence for Health promotes an ethical integration of AI into digital health care solutions based on six core principles [57], including discussing the transparency of AI in health care. Additionally, biased databases are problematic since they are often based primarily on male subjects and usually do not include minorities, possibly limiting the validity of the data [58,59].

Physicians, either practicing in primary care or a clinical environment, must still focus on compassionate healing as the foundation of health care, even though the mode of interacting may change. Frequently, continuity of care is impeded by insufficient data sharing among physicians and other HCPs (eg, information about treatment decisions). Simultaneously, being equipped with more time, opportunity, information, and technologies, the training of the empowered physician to treat and monitor remotely is also essential [17]. Furthermore, the voice of nursing and the viewpoint of frontline workers is described as a critical point in developing mHealth apps [18]. In this context, the empowerment of nurses and community health workers (CHWs) can positively transfer to their patients [9,18].

Various interactions take place in the health care network (see Figure 3). Within the network, individuals communicate, share data and knowledge, cooperate, or provide treatment and support [65,66]. While interacting, each individual has a specific role comprising their own skill sets, motives, and more personal characteristics. Additionally, the same individuals may interact in different ways based on their goal of interaction. Hence, the individuals’ roles and interactions may change depending on the underlying interaction type and intention (eg, primary care or research and development) [67,68]. Exemplarily, a physician might be treating a patient (primary care), whereas in another context they can contribute to research projects (development). While the primary aim relies on empowering the network of stakeholders, single nodes still retain decision power on downregulating or eliminating a specific interaction. This implies that individual nodes (stakeholders) and edges (empowerment means) can crosstalk in defining the final empowerment network.

Empowering the interaction does not merely mean reshaping or prioritizing relationships. Instead, such empowerment can be defined as identifying the therein present interactions, their underlying goals, and relevant actors, and determining how these interactions can be supported in a balanced manner. Given the multilayers and roles described above, the overall interaction network could end up being a complex interplay of intricate edges. Hence, we propose to approach interaction networks merely as a channel for information flow, independent of the types and number of participants. This allows for the formation of generalized constructs, which can be resolved in generating functional subnetworks for each empowerment task. For this purpose, we next aim at partitioning the health care network into functional subnetworks.

Putting these concepts into practice, in principle, almost all nodes of the health care network may interact with each other for multiple contexts and goals (see Figure 3). Following the idea of the communication flow, subnetworks can be identified based on specific tasks (see Figure 4). Subnetworks can overlap in part of their nodes and interactions, still yielding a different empowering task. In addition, one subnetwork may be entirely part of another subnetwork regarding their nodes and edges, yet aiming at another or more specific goal. Exemplarily, the primary care subnetwork and the continuity of care subnetwork share common nodes and edges, even if the final empowerment task is set to different aims.

Through pooling instances (ie, nodes and edges following the same goal), their weaknesses may be balanced out by the strengths and skills of other participating individuals. With this approach, the mentioned benefits of interaction empowerment are still incorporated, yet the count of empowerment targets is reduced to the count of chosen subnetworks. In summary, we treat the subnetwork as one instance following one goal with a set of strengths and weaknesses that may cancel each other out to a certain degree.

To provide practical use cases, we describe four chosen subnetworks, as depicted in Figure 4: clinical care, primary care, continuity of care, and research and development. To each subnetwork, we map the challenges of current health care that we assume to be positively influenced by it.

In the first subnetwork, patients, clinicians, nurses, and other HCPs interact for clinical health care. Here, lacking treatment adherence might obstruct successful clinical care. The empowerment of this network might help to tackle this challenge.

Similarly, the patient may interact with the family physician for primary care (ie, for day-to-day health care). Here, the lack of treatment adherence also poses a challenge. In this specific case, the network generally only comprises two nodes with one interaction entity. We nonetheless consider this interaction as relevant to form an individual subnetwork.

Third, the interaction among clinicians, the family physician, and the patient ensures continuity of care, as in the context of a hospital stay. However, unstandardized data formats and work practices, data security, insufficient intersectoral communication, and resulting incomplete medical records hinder the continuity of care. These challenges may, in our opinion, be tackled by empowering this subnetwork.

Finally, in research and development, researchers, developers, physicians (clinicians/family physicians), patients, and clinical administration may interact, with each node’s contribution and degree of involvement depending on the unique context/setting. Here, the researcher, developer, or clinician might present as the same individual, such as in the field of personalized medicine. Academia-industry cooperation and institutional review boards can also fall into this fourth category. Herein, the progress is slowed down by data security and privacy issues and insufficient intersectoral communication. Again, by empowering this subnetwork’s interactions, we argue that those challenges may be improved.

Some mHealth apps already support a concept of empower-oriented communication through interactions. In the following, we summarize some examples tackling different network empowerment concepts, which can be used as a starting point for further development.

CoCoV [69] is an mHealth app that supports patients to track their symptoms after receiving a COVID-19 vaccination through an easily understandable questionnaire. This app empowers the physician-patient interaction in preventing lost or overlooked symptoms. Additionally, the safely and anonymously transferred structured data can further be collected, enabling enhanced analysis of adverse symptoms by researchers and data scientists. Hence, the design of CoCoV simultaneously tackles two different empowerment subnetworks.

The app NEMO [69,70] supports patients receiving tumor therapy to protocol-related side effects in a structural and standardized manner. Typically, patients may forget to record incidents or may not consider them relevant [71]. The management app hence empowers the patient-physician interaction and communication in counteracting those issues. Since patients can administer the data sharing through QR codes, they can be empowered individually and continuity of care can be enhanced. NEMO [69,70] can be further used to empower the treating physician and other HPCs by facilitating data sharing (eg, among family physicians and specialists). Additionally, as with CoCoV, standardized and anonymized data collection empowers research and development.

Track Your Tinnitus is another symptom tracker enabling the patient to record daily tinnitus perception [72]; the app PainBuddy focuses on pain and symptom management of children with cancer [7]; and MindFrame empowers patients recently diagnosed with schizophrenia in managing symptoms, medication, and recurring behaviors [10]. Like CoCoV and NEMO, these apps empower the individual yet also support the interaction, communication, and data transfer, resulting in interaction empowerment of the respective subnetworks. Focusing more on empowered communication, Ramukumba et al [9] presented an mHealth app used by CHWs. Here, the support for communication through information exchange and timely decision support at the point of care resulted in the empowerment of CHWs, and consequently that of their patients.

FeelBack is a project assessing psycho-oncological stress through an app-based questionnaire. Here, not only patients can be targeted but also their relatives feeling burdened by the illness. Hence, by empowering the related individuals, the supportive interaction among patients and others can be empowered [19].

In simplifying intersectoral communication and data sharing, the browser-based system CTest enables tracking clinical samples online [73]. The tested individual can access their test sample’s status update through personalized QR codes. With this approach, clinical staff can be disburdened while the testee is not dependent on clinical staff for accessing test results. Hence, this interaction is empowered through simplification of the overall communication process.