Search

Modeling human behavior remains an immensely challenging task. However, such models are necessary. For example, to build completely driverless cars, the autonomous "AI driver" must be able to safely engage the human driver when necessary. The necessary AI-human engagement requires the AI to have a model of the human driver. Although a driving behavior is actually rather limited in comparison to the full scope of human behavior, it is still challenging to capture in a model, and is the subject of active research. Our projects and research indicate that we are developing a Human Digital Twin (HDT), which, of course, requires a model of the human it supports. However, we will begin with capturing simpler versions of human behavior: children's behavior! For example, in the case of the Patient-Human Digital Twin, the beginning is with Neonatology, where the measured parameters are limited in contrast to an adult patient, the medical history is considerably shorter, and the intervention possibilities are somewhat limited. As a result, we anticipate that the models will be simpler and more reliable.  The same concept underpins the TwinUp app, which predicts child abuse risk. Because children's habits, routines, and movement profiles are limited, the model necessary for the app to predict abuse is also limited, and thus the predictability is more reliable. With more data, use cases, and models, the goal is to gradually expand the Human Digital Twin initital models to capture more complex behaviors and eventually maturing them to capture more aspects of a person's life in the same way that children gradually develop their concepts and skills over many years.

One of the risks a child might face is abuse, which can happen in many different ways and in all different kinds of social groups, cultures, and countries. Every parent wants their kids to be safe and accordingly there are many apps that provide parents with various possibilities to connect to their children, manage their digital interactions or monitor their locations, or get alerts if their kids are out of specific zones. On the other hand, risk management and mitigation are heavily reliant on assigning a level of probability—a prediction—to the incidents that an organization wants to limit or avoid, such as accidents, losses, or project failures. In comparison, to provide successful child protection, a parent must gain insights into the risks they believe are relevant to their child and an early assessment of the child's level of risk exposure to mitigate risks before they occur. This is the primary goal of our TwinUp app: to learn a child's social interactions and routines, apply a community-specific abuse risk profile to predict a risk score to various child activities. Based on the score, the TwinUp app then can inform the child, parents, or caregivers of potential high-risk circumstances to mitigate the situation. In addition, the TwinUp app provides a digital network that connects a child's device to what we call "persons of trust" at the many places a child frequents to speed up mitigation and assistance when needed. It should be said that the TwinUp app doesn't need any data that other child protection apps don't already use.

Business intelligence (BI) provides insights on four levels: descriptive (what happened), diagnostic (why did happen), predictive (what will happen) and prescriptive (how to make it happen). The most crucial point to remember is that each level builds on the one before it. For example, a prescriptive task to determine the pricing for a service or product which would improve sales by 15%, is contingent on accurate sales volume predictions, which are based on knowledge (diagnostic) of what factors influence sales and consumer purchasing decisions, which is contingent on access to previous sales numbers. There is a challenge, however, with many intelligent mobile apps that prescribe actions to their users to reach some goals (for example, fitness, health or financial goals): they struggle with the second level, the diagnostic, and accordingly, their recommendations are not as effective as might be expected by the users. To "prescribe" training or diet regimes, a fitness or health app needs reliable diagnostic about why did a user achieve or miss their target. One way of providing this diagnostic is to rely on assumptions learned across multiple users which are then refined using user input via questionnaires. The problem with that technique is that it is tedious, and the users may be unaware of factors influencing their behavior and choices. Another technique for providing a more reliable diagnostic of users is to collect a broader range of user interactions across different aspects of their lives and rely on these interactions to learn user-specific patterns and preferences. Such user-specific patterns would reveal insights into the user's lifestyle and commitments, which might then be used by an app to enhance Its diagnostic and provide more effective recommendations. This is what the Information Digital Twin (IDT) is attempting to accomplish: observe user digital interactions, learn their specific routines, infer and classify their interaction preferences, and provide the user—and other apps, if desired—with a better insights Into the user's lifestyle and thus enable the users to achieving their goals more effectively. Our assumption is that many aspects of our lives are revealed through our digital interactions. Which app (or application) categories we use, how often, for how long, when, where, and in what order we use them—regardless of content—can reveal insights about a person's mood, lifestyle, and preferences, and enable accurate and personalized assumptions about why-type questions.

Enabling Human-Machine Teams Starts with Modelling their Interactions using a Standard Vocabulary-A Set of Information Tokens! Digital Twins for Human-Machine Teaming (HMT) Human-Machine Teaming (HMT) refers to situations in which people and intelligent agents, or robots, work together to achieve a common objective. It is projected that managing the increasing complexity of many domains such as aviation and smart cities can be only achieved through human and intelligent agent teams. Such teams would consist of-for example-pilots, air traffic controllers, and ground staff working together with aircraft agents, air traffic control agents, weather agents, maintenance agents, smart drones, and self-driving ground vehicles. In an effective HMT, and given a shared objective and task environment, each team member must be able to identify the actions it can take which will enable the entire team to accomplish the shared objective. HMT Challenge-Defining a Shared Task Model Each human team member, such as a sports team, has a mental model of the game's rules, the playing field, and their colleagues' roles (the task environment). Then, each player uses their own model to "simulate" numerous interaction scenarios and identify the ideal course of action for a specific circumstance. Even though each player follows the same game plan and objectives, if there mental models are not consistent (i.e., they have different interpretations of the game rules or the roles of other players) then it is highly unlikely that their actions are consistent with those of their teammates. Accordingly, the team will be ineffective in achieving their objectives. ​ Thus, the greatest obstacle to attaining effective HMT is providing each team member with a consistent model of the Task. In addition, the task model must be updated in real-time to reflect the task progress, thus allowing each team member to base their actions on the same task status, which increases the team's efficiency significantly. 00:00 / 00:55 00:00 / 01:25 The HMT Platform-Providing a Virtual Task Environment The primary objective of the HMT Platform is to provide each team member with an up-to-date model of the task they support. Each team member can then use the model to choose which actions to take based on their role. Each team member's actions, in turn, update the task reference model and accordingly influence the following decisions of all other team members. The task reference model is the Information Heat Map (IHM) . The IHM is a data model representing a task (or scenario) in terms of the many input and output parameters defining it, their correlations, and conditional dependencies. Using the IHM as the environment, each team member can then use reinforcement learning to determine the optimal course of action for attaining the team's objectives. The IHM thus provides a virtual task environment, the task reference IHM, in which all team members communicate and are connected in real-time, thus enabling an effective HMT. IHM_HMT How is the HMT Platform Different? The HMT Platform is initiated by defining the scenarios capturing the various tasks as information models, the Information Heat Maps, IHMs . A member's unique IHM combines one or more Reference IHMs to cover the asks the member needs to support. ​ The IHM's most significant innovation-and difference-is capturing scenario parameters as a range of discrete values. For example, temperature is recorded in discrete value ranges: 35-37, 38-40, or > 40. A measured value is then represented as a "1" for the matching range and a "0" for other values. ​ The discrete representation of data enables the HMT Platform to capture, describe, and correlate any metric type in a single, consistent structure: the IHM. In other words, an IHM can model and represent a multitude of complex tasks or scenarios as a single-task representation . Team members can then decide what to do based on the same, up-to-date representation, which they can access through their own apps (HDT/IDT).. 00:00 / 00:58 00:00 / 01:33 Why is the HMT Platform a Better Solution? The HMT Platform and the connected HDT/IDT Apps create a virtual task environment similar to that of multiplayer online games. According to this analogy, a team member's HDT/IDT stands for the avatar with which they interact with the environment . In addition, and as in online games, only that part of the environment is rendered which is relevant to a player, i.e., each team member's IHM represents only that part of the environment that is relevant to their role or task. . ​ On the other hand, scenario and task owners can manage a complete team's tasks by modifying the parameters of the relevant IHMs, which then propagates through all relevant HDTs/IDTs and accordingly updates their subsequent decisions. ​ Finally, mature, role-specific HDTs/IDTs can be assigned to new team members to speed their learning and adaptation to their roles, ensuring task performance consistency. Building Human-Machine Teams-How to Start? As with other HDT/IDT use cases, building digital twins-enabled HMT environment starts by identifying a set of tasks and scenarios to be carried by the HMT team and then build a prototype of the various involved roles to ensure the feasibility of the HMT Platform. Please refer to our services page for more details! TECHNOLOGY

Patient-Centric Predictive Solutions Start with Establishing a Common Vocabulary for Medical AI! Patient-Specific Predictive Diagnostic Even tiny changes in patient vital signs and measured values can signal a developing problem. Time is of the essence when it comes to predicting these deviations , placing them in context, understanding them, and reacting to them with a therapeutic change. These tasks are especially demanding for physicians who have little experience with the patient group. Challenges with ICU Care No reliable technology for predicting ICU patients' critical events No patient-specific predictions or risk alarms Critical ICU events may have long-term effects on patients Patient risk assessment is still dependent on the attending physician's competence and knowledge, which may vary between shifts and facilities Many critical events may be predicted if there is real-time, comprehensive patient monitoring Longer Mean Time to Respond (MTTR) after identifying a critical event (assessment, data analysis, treatment option assessments) raises patient risk High treatment cost and limited ICU capacities The Patient-Human Digital Twin (Patient-HDT) Provides automated and patient-specific predictive risk notifications Provides treatment recommendations and simulations Provides consistent and expandable knowledge base to close the knowledge gap across medical teams Compensates for the increasing lack of specialized medical personnel Expandable to cover various types of patient metrics and treatments Open architecture that can integrate with, and learn from any patient-related data source How is the Patient-HDT App Different? Patient-specific, c an be provided as a service to patients to improve risk prediction after they return to their everyday life Continuous learning from patients’ digital interactions to enrich risk assessment and recommendations Continuous and automatic updating with new medical information to improve risk predictions and mitigation recommendations Insights from diverse Patient-HDT can be incorporated into a single dynamic medical knowledge base to use anywhere to enhance treatment quality Start with the Patient-HDT Feasibility Assessment Building data-driven healthcare solutions is a challenge. Not only because of patient data privacy, but also because of the extensive specialized medical expertise required to engineer this data and provides insights from it. Our method uses raw data to define initial data-related insights to determine the feasibility and value of the Patient-HDT. ​ The first step is identifying a group of critical conditions or risks that significantly impact treatment. Next, a set of parameters and metrics associated with the conditions or risk is identified. The data associated with the parameters and metrics in scope is then obtained from historical health records. The Patient-HDT uses the obtained data to learn the information model particular to the conditions or risks under consideration (Information Heat Map, IHM ). The IHM provides initial insights into the correlations of the factors impacting a condition. Further, the IHM allows the medical team to gain insights into how controlling specific parameters would impact the condition or risk of a patient. ​ Validating such insights is the first step towards validating the value and feasibility of the Patient-HDT! Events Risk Profile Example Modeling Patients, A Key Challenge for AI Solutions in Healthcare Digital twins are expected to offer unprecedented insights into a variety of fields of medicine and healthcare. At some point, digital twins will fundamentally alter how physicians engage with their patients. Doctors would employ digital twins to diagnose patients' conditions more accurately, simulate treatment possibilities, and provide assistance during life-saving surgery. ​ On the other hand, the human body is an extraordinarily complex and dynamic system. Constructing a model of an organ or a patient remains a complex undertaking, made even more challenging due to the sensitivity around patients' health data. ​ However, utilizing the Patient-HDT Platform enables a relatively simple and staged introduction of digital twins into the medical sector. For example, the radar chart depicted here, which models a hypothetical medical risk event, can provide preliminary insights into the significance of several medical and lifestyle factors associated with that risk event and their likelihood of contributing to that event, all without application of elaborate machine learning techniques. ​ Once the model have been confirmed for its consistency, accuracy, and value in terms of risk reduction, the more elaborate implementation of the Patient-HDTs for the individual patients can commence. Patient-HDT Data Structure The patient Information Heat Map (IHM) , which captures all pertinent patient and treatment data, is the central component of the Patient-HDT. The Patient IHM is initiated using historical patient data. A cell in the heat map depicts, for instance, the degree or association between a particular dosage of a particular medicine and a particular heart rate range. Using Bayesian inference and the same set of historical data, the Patient IHM determines, for a given risk event (any combination of output values), the correlation of each input on that risk event. These values are then used by the Patient-HDT to build the Event risk profile, to determine the onset of critical events and recommend courses of actions to mitigate the risk. PATIENT HDT WHITE PAPER

What is Information? What is Intelligence? Building Artificial General Intelligence (AGI) Agents We developed the Social AI Platform and the Information Digital Twin (IDT) to provide commercial AGI capabilities and solutions. As such, the Social AI Platform is intended to support a wide range of intelligent scenarios using the same agent: the Information Digital Twin (IDT). ​ The Social AI Platform technology is the realization of our novel concepts and approaches to information and intelligence. In its mature form, the IDT can be considered as a general reinforcement learning (RL) agent Information and Intelligence-Unified Definitions* We believe that the path towards building Artificial General Intelligence (AGI) agents is based on unified-and dependent-definitions of information and intelligence ​ Artificial approaches to intelligence depend on computational models to process information and provide intelligent capabilities. Due to the lack of uniform definitions of what constitutes intelligence and what is information, the capabilities such models provide differ according to their interpretations of intelligence and implicit assumptions about what is information. The variety of interpretations of intelligence and information also indicate that existing intelligence computational models provide specialized rather than general capabilities. The following are the unified definitions we have developed so far and use to mature the IDT and the Social AI Platform. Human-Machine Teaming (HMT) is the prominent study area for unified information and intelligence definitions, focusing on enabling intelligent collaboration across fundamentally distinct entities. Hafez, W., (2022) Information as Entanglement—A Framework for Artificial General Intelligence, In: Goertzel, B., et. al. (eds) Artificial General Intelligence. AGI 2022. Lecture Notes in Computer Science, vol 13539, pp 20-29, Springer. DOI: 10.1007/978-3-031-19907-3_3 What_is_info Information A system's degree of entanglement with its environment measured in bits. That is, its capacity to predict its environment and be predictable by it Information (and intelligence) are apparent through interactions. If an agent or a system does not interact with its environment, it is impossible to tell if it processes information or if it is intelligent. Interaction, or more precisely communication, is thus a prerequisite for understanding and defining concepts like information and intelligence. ​ To communicate, an agent relies on a set of symbols, alphabet, or code and some rules to construct messages and to send them to its environment, and to deconstruct the messages it receives from other agents. ​ If the various agents depend on one another in reaching their goals, then the set of symbols used by each agent to build its messages will start developing some degree of correlations. Although each agent might use a completely arbitrary set of symbols, they need to make sure to use these symbols consistently in order to convey their messages effectively. This consistency is what achieves the correlations among the various sets of symbols. ​ Once correlations are established, predictability follows. When an agent sends a specific message, it predicts receiving a particular return message. In the other direction, when an agent receives a message, the level of correspondence of its response message determines the environment's capacity to predict its behavior. Our research assumption is that the level of correlations and the resulting agent-environment predictability is a unique, general property of a system interacting with its environment, similar to properties such as energy or mass. Consequently, we attempt to demonstrate that information is the metric for measuring this level of correlations, or entanglement . To formalize the information metric, we then rely on concepts from communication theory. Intelligence The activities of a system to maximize its entanglement with its environment, That is, the system strive to maximize its information, and accordingly its capacity to predict and be predictable The relationship between intelligence and information, or level of predictability as we define it here, is rooted in the nature of communication. Communication theory (Shannon and Weaver, 1949) defines communication as "the procedures by which one mind may affect another." To include "non-mental" systems, communication is thus the procedure by which one entity influences the actions of another in a particular direction. An example is a thermostat sending a control signal to a heater to change its degree of transferring energy into heat. Developing an effective communication system is a highly complex task. An agent needs to define a limited set of unique symbols, define rules to structure them into signals and messages (syntax), and map them to the set of concepts necessary to achieve its objectives (semantics). However, the most challenging aspect of communication is the activity of maintaining, updating, and optimizing the symbols, syntax, and semantics rules to achieve effective communication. This aspect is where our research assumption about the nature of intelligence comes into play. Our research assumption is that intelligence is the continuous activity of an agent to evaluate its communication effectiveness and increase its capacity to influence its environment toward achieving its objectives. This translates into monitoring the usage of its communication symbols, updating the rules to build messages out of these symbols, updating the messages to reflect its objectives, and ultimately evaluating its overall success in influencing its environment or adapting to it. ​ The question here is, what is the control signal that helps the system to accomplish these activities? Our assumption is that one possible way for the system to measure its effectiveness in completing these tasks is its overall information: its ability to predict and be predictable. That is, the intelligence of an agent can be assessed by its predictability capacity and if it can maintain it or increase it against changes in its environment. Research Opportunities: Integrating & Managing Heterogeneous Systems One method to substantiate the definition of information as a measure of entanglement is to demonstrate how to use information as entanglement to i ntegrate, control, automate, and optimize heterogeneous processes that span systems of fundamentally different natures (humans, machines, smart agents as in the case of Human Machine Teaming ). The Information Digital Twin, IDT (and its human-specific counterpart, the Human Digital Twin, HDT ) enables there capabilities by relying on capturing their level of dependencies: i.e., information. We thus seek research and partnership opportunities to expand and mature the concept and application of the IDT/HDT. ​ A typical output of such research is an IDT/HDT prototype that automates and optimizes heterogeneous processes based on an information model (the Information Heat Map, IHM ) that captures the correlations between the parameters of the various systems, notwithstanding their nature. Please contact us for more details, research papers, or trial algorithms Realizing General Intelligent Solutions: The Information Digital Twin (IDT) & The Social AI Platform The Information or Human Digital Twins (IDT/HDT) are the technical implementation of the unified information and intelligence concepts. The IDT/HDT architecture and algorithms enable them to assist their users in measuring and assessing the level of information (entanglement) involved in their interactions with their environment and help them optimize their actions towards increasing their information (i.e., entanglement) over time. Entanglement, Predictability & Information Our assumptions concerning the unified definitions of information and intelligence are founded on the concept of entanglement, a fundamental characteristic in quantum mechanics. According to Erwin Schrödinger, 1948, "entanglement consists in the fact that a single observable (or set of commuting observables) of one system is uniquely determined by a single observable (or set of commuting observables) of the other " For two entangled systems, say A and B, knowing the state of one of its parameters provides system A with complete knowledge (prediction!) of the corresponding parameter in system B. The higher the entanglement between the two systems, the higher their ability to predict and influence one another. ​ As we propose, information is thus an indication and a measure of the level of entanglement between two systems, which corresponds directly to their ability to predict and be predictable. Entanglemnt How the IDT uses Information to Enable Intelligence As an AGI agent , the Information Digital Twin, IDT, or the human-specific version, the Human Digital Twin, HDT, enable their users to measure their level of information (entanglement or predictability) as they interact with their environment. This measurement is based on calculating the probabilities of the various parameters involved in the user-environment interactions as captured in the Information Heat Map (IHM) . The various use cases (Shipment Digital Twin , Human-Machine Teaming ) provide examples of the information map structure used to calculate the user-environment information. As indicated in the two examples, any change of any involved probability would change the value of the user-environment information. The IDT can then use the information change to update its search for alternative actions to maintain or increase the user-environment information, which reflects the IDT intelligence. Currently, we have two pending patents covering the Social AI Platform's (and the Information Digital Twin) novel techniques, architecture, and algorithms Research Focus-Improving RL Agents Performance and Adaptability Current RL (reinforcement learning) agents rely on an interpreter* (or the agent designer) to establish the task states, rewards associated with different states, and agent objectives (e.g., end state). Changes in the environment and/or the agent's objectives necessitate the interpreter to intervene and adjust the states and/or rewards to adjust the agent's behavior accordingly. Suppose the agent can independently update its states and rewards based on its objective. In that case, it will become more adaptable to changes in the task environment and more general because it can adjust its behavior to multiple objectives. ​ * https://en.wikipedia.org/wiki/Reinforcement_learning General RL Agent Using Agent Information (its Level of Entanglement with the Environment!) to Optimize its Actions Replacing the observer with the Information Heat Map (IHM) , which captures agent-environment interaction information, will allow the agent to dynamically update its states and rewards based on its objectives and thus be able to adapt to changes in the environment. ​ The current emphasis of our research is to validate this assumption. Furthermore, suppose we assume that the agent's ultimate objective is to improve its total information. In that case, the agent can use changes in interaction information to simulate numerous actions and choose the ones that will increase its information in the long run. INFORMATION DIGITAL TWIN