4. Method
Method
Introduction
This method proposes significant improvements in the fields of preventive medicine and Healthcare management. Then, by extension, of all areas of the digitally mediated human experience.
It does this by enabling the formation of collaborative predictive modelling networks corroborating valid relationships between a person’s individual traits and characteristics (the “quanta”) and their relationship to future events stored in existing databases as past-events of people found to be most-like you, under circumstances most-similar to yours.
This paper cannot provide the method for this to be executed locally, as this is entirely dependent on local database structure and content, but basically means: it recognises what types of behaviours tend to be associated to what type of movement/behavioural patterns as expressed by others.
Practically speaking those “past-future events” (past for them, future for you) which can be offered for comparison, are those of other, “most-like” individuals. This, in effect requires the local construction of large learning models (“archetypes”), an industry in full swing. With these behavioural avatars, the clinical world can open up access to a seismic shift in improved user-feedback loops, increased efficacy of care pathways and cost-efficiency of the health of nations.
The proposal in this paper only concerns itself with disseminating the idea itself and aiding access to the analytical power inherent to it. First, the power of comparing an individual’s health and their physical behaviour to that of an avatar with “most-similar” amalgamation of traits and behaviours must be established. Then, the power in the distribution of insights associated to those most-similar individual averages as clinical advice to patient and clinician alike.
Later, other aspects of comfort, wellbeing and productivity in the form of improved personalisation and optimisation naturally emerge. In effect, this system does statistical research in real-time on questions asked with information given.
A search tool that evaluates what would be best for you as demonstrated by people “most like you”.
Creating the avatar – Step One
The first step is to design a digital avatar that represents a given individual’s real-world experience, in a virtual 3Dimensional matrix model. This avatar can be made from a Lidar-scanning process integrated within a CCTV-solution as part of standard privacy laws.
It can even be started with standard CCTV solutions, but regardless offers digital representation of a movement pattern with geo-location qualities. Very basic data such as stride-length variability or shoulder/hip sway can then be computed as a “real-world movement pattern”. As such, a movement pattern can then be geolocated and corroborated for deviation/similarity to others anonymously, creating normalised archetypes. Individuals can then be considered as “deviating from the norms associated to people most-like them, under circumstances most-like theirs, to a greater or lesser but quantised degree”.
Quantised. i.e. There is no such patient who is “normal” in all-traits, but there is such a thing as a patient who appears “within normal ranges, when considered from a limited number of chosen perspectives, circumstances and descriptors (language)”.
“Normal”, after all, is only a time-limited, averaged, calculation of multiple and widely varying behaviours, traits and characteristics. These, on evaluation of their overall appearance, seem to (at least from the perspective of the observer) be within the deviation-range that said observer is willing to consider themselves to still perceive as operating within acceptable societal boundaries when taking into consideration context offered.
What is “normal” if not a “thing” or a “possible state” rather than an actual one? It is in essence an evaluatory function composed of a series of partial truths being held together as if representative of the complete truth. An almost-truth. Full of “emergence-potential”, yet an entirely hypothetical, illusory combination of possible average states designed to steer us in the right direction.
A wave-particle. The idea that an individual could both “be” and be “normal”, is technically absurd. Their blood pressure, however, (or some other component that serves to describe part of the whole) can be within normal range for a class of patient. Something as obvious as stating that “there is someone out there for whom your normal isn’t normal” doesn’t require further expansion.
What this paper draws attention to is realising the potential of what normative data can mean.
It removes the theoretical boundary of evaluating pools of patients against cohorts of non-arbitrarily (statistically confident) pooled patient-data, all classified by clinical observation (measurement error).
Representing the individual, not as a privacy file but in the form of AI-question-based avatars for easy calculation, in turn offers a computational perspective that represents behavioural data as the key-definers of the archetype. Not symptomatic data as is currently largely the case.
The reasons for how they feel/look/behave, not the outcome. The symptomatic and private-identifier data remain within their respective local databases. The primary data point for this integration is a change in movement pattern that can be observed within the consistency of the individual’s rhythm, shape and mode of propulsion over calibrated distances and surfaces.
Image of surface topology of an avatar of an X-Y axis going through a star-jump. Changes before and after treatment can be analysed and corroborated to symptomatic changes.
This can be done with current technology by creating a digital avatar of some dynamic signature that has real-world value (i.e. was done in the Real-world) and attaching data-labels. This can for example be achieved using the Nowak-Sitnik Reference 1. method of 3Dimensional motion rendering of human shapes. Something non-intrusive and the most-distant snapshot (yet valid)-evaluation of another person’s “life quality”: Walking. This works from the micro to the macroscopic with the obvious observation of a symmetrical vs limping gait indicative of some expectation of trauma, to subtle changes in stride-length and posture, possibly in response to therapeutic encounters or changes in mood.
The question of whether this offers proof of effectiveness of individual therapies is irrelevant to this paper, but merits discussion and is found in Addendum A.
This avatar is representative of a bundle of patterns, associated to bundles of other patterns, each in turn representing statistically similar people who did similar things and behaved in similar ways to achieve desirable outcomes under similar circumstances. This then, equally logically, offers the ability to cross reference the cluster value found across multiple most-like individuals’ avatars, against representative ethnographic and epidemiological data. In the process, this exponentially increases its predictive capability with each additional data point.
(Image showing a clinician in the right lower quadrant pointing a patient’s attention to the left upper quadrant where a person who did what the person in the left lower quadrant is being suggested here, has done)
These patterns, devoid of any personal identification, have already been widely studied and observed, and will function as reference base for the inherent confidence of their individual deviation-from-norm. In this fashion, the concept of “normal” as an average representation of multiple values, traits and characteristics, is now being quantised in clusters of smaller packets of micro-behaviours. This, in a sense, is the natural progression on what we have always done and formalised as the scientific process; we noticed flushed cheeks, measured a blood pressure, and found out about cholesterol. Identifying ever-smaller packets of data within larger clusters of data, observing their relationships, weighing up their impact and value, and recording it for future repetition and analysis to come to the more-meaningful conclusion and prediction is the very essence of knowledge, science and prediction.
In Infodynamic terms; we are managing the entropy of the system and observing the transformation of high-entropy patterns to calculationally more agile low-entropy patterns modulated by rules and pursued by behaviours. In this individual-centric (observer-centric) computation, the concept of “normal” is: “that which the patient defines as more desirable”. From this point of view we can conceptualise it as incorporating a measure of the biased individual perspective on the mathematical framework. This offers algorithmic understanding of the changes in the entropy, as set and defined by the individual’s perspective and satisfactory results are achieved when those values are met. A benefit of this strategy is that even after the “data-exit of the individual” (removal or loss of the key to their wallet) from the AI-generated search question, the interaction with the data and the archetypes held within the data have now been modified as a networked result of that interaction and the system has automatically and permanently become more intelligent.
These bundles of patterns that link the digital twin avatar to the clinical file, only associate to traits and features whilst all private data remain in the local database. The anonymously emerging behavioural patterns, then form archetypes that can be processed centrally, and analysed for further emerging patterns across wider population groups.
By recording and distributing these avatars more widely geographically, new coalescing and emerging patterns can be seen. Something we have already seen throughout the history of modern medicine and defines the scientific process.
The technological innovations underpinning this are minimal and usually already in existence. Visualisation equipment, algorithms, database management tools, calculational tools from the financial sector and mathematical relations from Thermodynamics, all readily available. Neural Networks, Large Language Modelling and blockchain ledger systems all exist today. With the addition of some calibration, existing CCTV and autonomous clinical, healthcare and educational governance, data systems will be able to explore this functionality widely to their users' benefit.
Step Two
The second step is to take the identified movement pattern (e.g. “Stride-length variability if identified by CCTV or Lidar) as one of a limited number of subcategories of arbitrarily weighed statistical categories, populated as presenting (randomly) at a location and identified by pattern name for that event, at that moment in time. (it is impossible for two people to be in the same place at the same time, and therefore perfect randomisation). For integrative and clarification purposes, we make a naming suggestion for this randomisation process as named after the platonic solids. More discussion on Platonic shapes and “Platonics” refer to Addendum C.
With this step, an arbitrary data allocation is given to a recorded “event” and is categorised as (a series of) measurements over a given (limited) time span at a given (permanent) location. It is from now on impossible for that data-point's locale and timing to be altered or deemed false. This is significant in AI and cryptography and discussed in Addendum E but not the topic of this paper.
The point is that having been recorded as an event at that location, no other event of the same class could have occurred in the same place simultaneously (two tennis balls cannot be at the same point in space and time during a match).
This could be assessed using CCTV over a hallway's distance and a treatment visit's duration. The allocation of an individual’s change in shape of motion in space (suggested term: Kinesio-morphological topology) as a distinct and calculable item within a subcategory, allows for people who are technically moving in the “same” way elsewhere, to be allocated as different (as being of a different subgroup) calculationally. A difference based on, and produced by, their allocation to a location in time by random distribution.
Side note: It is somewhat ironic, poetic and yet simultaneously unsurprising that the individual, the thing that can most efficiently communicate with us consciously (and of which there will only ever be one to be found of in any one place/geolocation at any one time) would be the observer perspective to provide this AGI computational facility.
Perceiving order:
There are multiple computational levels running simultaneously in the process of observing “changes in shape of behaviour(-al topology) in response to the surrounds” (Topological Kinesiomorphology or TKM). These can appear rather random from some perspectives but with growing knowledge in healthcare and understanding of the interrelation between psychology, physiology and physical health, we are starting to see order appear as those changes are now becoming understood as being constructed from numerous events which run parallel computational processes exploiting the same data-lake from differing computational perspectives to coalesce into that observed moment. Meaning:
Here we refer to them as first, second, third etc. order events.
A first order event is delineated/defined by the physical limitations (battery-life/measurement accuracy) of the measurement device itself. As such, for example, a first order event for a static (location-based) camera is limited by the lenses' reach and the sensitivity or lifespan of the battery for the heart rate-wearable.
Second order events are evaluation of recurring trends observed in first order events (avatars). These different classes are naturally fundamentally different because their computational perspectives differ.
Third order events are patterns observed on interactions between second order events of different types (Traits) with this process repeated indefinitely and into laterally associated Quantae of data (movement pattern and height, BP, pain location etc., etc.).
The specific point is that reliable evaluation of a tremendously complex and networked event like “an Individual’s Health”, must, in its very essence, contain at least a dataset that describes how they move in physical space. This concept is often embodied in clinical notions of motion disorders or posture, but this is in effect, and less pathologically (and more generally), a coalescence of micro-expressions of motor-sensory responses that become classified and interpreted as constituting “human behaviour”. We discuss poor posture of movement “dis-order” without reference of relative “movement order” for that individual. We often don’t know what is normal for you.
Therefore, our proposal quite naturally insists that:
For accurate evaluation and prediction of something as complex as a person’s “Health”, it must also contain representation of the “true physical localisation of human movement-variability as response to external stimuli”.
This would be understood as obvious when observing that human behaviour is largely, even if not solely, a coalescence of automated and habituated sensory-motor responses.
In understanding the physical response to the environment and the physical reaction to physical stimuli such as healthcare processes, we can create a true and intimate understanding of the intimate detail of neural habituation and health conditioning of an individual.
For analysis of a system to achieve this, it must, by necessity, also be measured in visual terms from a fixed-point-location, offering geo-location and timestamping to the mapping of the data that locates a person’s body in a point in space (localisation).
This can easily be achieved at a standard commercial site with a specific interest such as a healthcare clinic, hospital site or assisted living facility.
Step Three
The third step is to attach the data representative of the individual (The surface avatar) collected from third order events to non-personal identifiers like areas of pain, diagnosis and treatment protocol in the most effective way. This is done locally as a clinical database system with topological surface data and health-stats statistical variance and trends. Surface-mesh changes are evaluated in clusters of patient groups for relevance and attributed as one of the platonics by highest-order event available. A mesh is constructed from the patient data, and again a sequence of individual trait-based 5-order sub-clustering is achieved by standardisation and distribution as in one of 5 archetypes. These sequences are evaluated and attributed as descriptor-data, essentially as if an encoded DNA sequence of that event represented as averages of the overall presences and behaviours (the group to which they analytically belong).
Computationally this is mimicked by n-Ary tree logic on functions. Logistically these functions appear as AI-based pattern recognition with trained LLM/ neural networks.
There is logically no need to use all 5 shapes of the Platonic shape symbology. In theory 3 would be sufficient as they would create the necessary randomisation. Whilst this is true and should suffice, interesting simulations of the aesthetic of that randomisation yield the question whether a relationship can be established across multiple numbers of these arbitrary subcategories and not perhaps only 5 either but this discussion does not form part of this paper.
With this classification of the meta-meta data, a novel computational perspective is enabled.
Closing the loop
It is easy to see the value of this proposal if an individual’s position and momentum were truly and fully “known” as part of a “closed system”. This would be quite impossible to do perfectly, but with lots of threads linking our current experience to that of thousands of similar people’s past-but-similar behaviours, we can begin to make a confident prediction. In this fashion, we can identify other “most-like” past-potential trajectories (treatments received/outcomes achieved) and select for the ones with the best (most desirable) associated outcomes. Then, over time, increase the strength of their relevance to the individual in front of the clinician by evaluating the trend lines yet again.
By establishing such an avatar-to-archetype link, significant re-evaluation of the available data can occur, offering significantly enhanced decision-making. In clinical real-world terms, this is lives.
Further interactions sought and received through apps can then be integrated into the pattern-evaluation warehouse via API’s. This in turn inevitably improves the theoretical value of the archetypes on their every interaction with new data. For further discussion on theoretical application of the data super-asymmetry please refer to Addendum B. where the geometric/ mathematical relationship between the different data types is expanded on.
Refinement
Stepping further forward into the field of evaluation of human behavioural patterns (Topological Kinesiomorphology) we can find significant scope for detailing and lateral extrapolation:
Acquisition of more detailed micro-behavioural data (facial expression /breathing /HRV/Pupillary dilation/thermal distribution etc.) provide further data on potential correlations to mood, anxiety, or responsiveness to stimulants. This can easily be acquired through user-interface devices for simulation and application of improved healthcare interfaces. This, both digitally and in real-world terms through improved educational adaptation and integration with wearable devices and new sensor technology. This can then also further be integrated into fall, injury or disease-prevention and early detection of neurodegenerative disorders, educational and perceptual challenges. We are already familiar with the predictability of certain cardiovascular and neurodegenerative disorders based on trends like heart rate or stride length-variability. Further extensions of this analytical logic in all the service fields of healthcare are inevitable. At its most superficial, it is easy to imagine this as part-basis of diagnostic and cost-optimising tools in hospital corridors or as preventive systems in senior/assisted-living.
This proposal is the backbone of a real system of evidence-based medicine wherein environmental cofactors as well as behavioural trends are solid predictive datapoints within the analysis of the patient. With further iterations, significant improvements to care and savings on cost can be made representing rapid, welcome and necessary improvements. This, whilst largely using existing infrastructure and without disrupting current economic markets and systems.
In infodynamic terms, this proposal creates quantised fields of associated locations in space and time. All in linear relations and each location tagged for easy retrieval in the form of anonymous archetypes. Patient X, filling the CCTV camera’s location at moment Y, does so displaying a motion pattern that can be described as a mathematical function over period Y-Z. In essence, a digital avatar is composed from the changes in motion-pattern.
These pattern-shapes and changes in shapes, are typical of certain physical states and brain/personality types/classes (e.g. neurodiversity) creating locality to an individual within the database’s 3Dimensional geo-location mapping. Patient X’ avatar is the equivalent of the particle in Shannon entropy terms and their existence leaves a visible trail of data of various natures which appear to coalesce into what we describe as their “life-experience” or experience of life, at the instant of evaluation of the data in the medical file.
