The Intelligent Phenotypic Plasticity Platform (IP3) for Precision Medicine-Based Injury Prevention in Sport

Abstract

The best predictor of future injury is previous injury and this has not changed in a quarter century despite the introduction of evidence-based medicine and associated revisions to post-injury treatment and care. Nearly nine million sports-related injuries occur annually, and the majority of these require medical intervention prior to clearance for the athlete to return to play (RTP). Regardless of formal care, these athletes remain two to four times more likely to suffer a second injury for several years after RTP. In the case of children and young adults, this sets them up for a lifetime of negative health outcomes. Thus, the initial injury is the tipping point for a post-injury cascade of negative sequelae exposing athletes to more physical and psychological pain, higher medical costs, and greater risk of severe long-term negative health throughout their life. This chapter details the technologies and method that make up the automated Intelligent Phenotypic Plasticity Platform (IP³)-a revolutionary new approach to the current standard of post-injury care that identifies and targets deficits that underly second injury risk in sport. IP³ capitalizes on the biological concept of phenotypic plasticity (PP) to quantify an athlete's functional adaptability across different performance environments, and it is implemented in two distinct steps: (1) phenomic profiling and (2) precision treatment. Phenomic profiling indexes the fitness and subsequent phenotypic plasticity of an individual athlete, which drives the personalization of the precision treatment step. IP³ leverages mixed-reality technologies to present true-to-life environments that test the athlete's ability to adapt to dynamic stressors. The athlete's phenotypic plasticity profile is then used to drive a precision treatment that systematically stresses the athlete, via a combination of behavioral-based and genetic fuzzy system models, to optimally enhance the athlete's functional adaptability. IP³ is computationally light-weight and, through the integration with mixed-reality technologies, promotes real-time prediction, responsiveness, and adaptation. It is also the first ever phenotypic plasticity-based precision medicine platform, and the first precision sports medicine platform of any kind.

Keywords: Fitness; Genetic fuzzy systems; Injury prevention; Mixed-reality; Musculoskeletal injury; Phenomics; Phenotypic plasticity; Precision medicine.

Figure 1.

An example of a stress-response curve. Note the areas of growth and decay on opposite sides of the peak response region of the curve. The dotted line indicates the fitness response relative to a “no-stress” condition.

Figure 2.

Two examples of ordinal fitness curves. The left curve is an example of a gradual increase to a peak fitness response, followed by an equally gradual decay. The right curve is an example of a more gradual increase to a lower overall peak fitness response (compared to the left example). Following the peak response there is a rapid decay of fitness. In comparison, the system on the right did not achieve as high of a peak fitness response and was overwhelmed by stress sooner, with a more rapid decline in fitness, compared to the system on the left.

Figure 3.

A schematic of the IP³ system. External configuration files form the foundational component of the logic system that drives progression through IP³. This component is responsible for deploying Unity 3D modules that, in turn, are the building blocks for each mixed reality scenario, including both the performance task context and the initial positions and subsequent behaviors of all non-player characters (NPCs) present in a given scenario—see Section 3.1.A. Behavioral profiles and summary variables are computed and processed (see Section 3.1). Phenotypic plasticity curves (see Section 3.1.C) are computed and submitted to the genetic fuzzy system global (GFS-G) model (Section 3.3.B) to determine the between-trial loading adaptations. Likewise, summary variables and frame-by-frame data are fed into the genetic fuzzy system local (GFS-L) model (Section 3.3.A) to personalize the within-trial behaviors of NPCs.

Figure 4.

Schematic showing interactions between the various scripting components, data streams, and GFS models. External configuration data are passed to the Trial Manager component which implements a given configuration by dictating the initial conditions of the Goal Translator and the opponent NPCs. An Environmental Observer component has complete knowledge of the scene and generates a representation of the state of the virtual environment at each time step, including the positions, orientations, and velocities of the athlete, the goal NPC and all opponent NPCs. In order for the NPCs to respond appropriately in the environment, the NPCs State Machine component polls the environmental state data structure from the Observer. The Observer is polled at a variable rate to control the responsiveness of the athlete, and individual subsystems drive the NPC behavior for a given trial (see Section 3.4). At each time step, the environment state and all data from peripheral devices (i.e., devices utilized to capture neurophysiological data, physiological effort data, and pose estimations for kinematic data (see Sections 3.1 and 3.5) are passed to the Data Writer component, where they are collated, time stamped, and exported to an external file. Once a given session is completed, these data are synced with the WrnchAI pose estimation solver for analysis. The resultant information from the analysis is used as the input for GFS-G. The product of the GFS-G model is then used to modify the external configuration for subsequent trials.

Figure 5.

Schematic of portable camera configuration. Each camera emplacement consists of 2 cameras on a single tripod, connected via 10m USB-C cables to a controlling laptop. Laptops calibrate to the same origin, creating a single 16 × 10m capture volume in which the virtual scenario is placed. See Notes, Section 4.8 for a description of each piece of equipment represented in this figure.

Figure 6.

Bottom row – Initial conditions for an example three NPC trial from three perspectives: (1) First person view of the participant at the start of the trial (left), (2) third person view from behind the participant (middle), and (3) top down view (right). Middle row –Trial conditions at approximately 2000 ms into the trial, with left, middle and right panels providing the same views as described above. Top row – Trial conditions at approximately 4000 ms into the trial, with left, middle and right panels providing the same views as described above. The variable, β, indicates the bearing angle at each time, with a greater difference in β from t1 through t3 indicating less efficient interception performance of the NPC target by the human participant.

Figure 7.

Schematic diagram illustrating interrelated classes and data structures that are saved for each trial. The DataWriter class collates per-trial initial condition data (a TrialConfiguration) and a list of per-timestep DataFrames. A DataFrame is a timestamped container for EnvironmentState data and PeripheralData from connected devices. The EnvironmentState for a given frame is a collection of EnvironmentObjects, which possess descriptors of position, heading, and object type - Goal, NPC, or Agent (the participant). NPC EnvironmentObjects possess additional descriptors, including Steering Dynamics Model data, Motion Matching animation parameters, and current state machine behaviors. All of this data is exported as a JSON, which is a hierarchical data format that maintains the relationships between these data structures.

Figure 8.

An illustration of our factor analysis design. GF (Global Fitness) is a second order latent variable, with N_Ff (Neurophysiological Fitness factor), O_Ff (Oculomotor Fitness factor), BFf (Biomechanical Fitness factor), E_Ff (Effort Fitness factor) and L_Ff (Locomotor Fitness function) as latent variables (i.e., dimensions). Each square below the latent variables symbolizes a specific measured variable within each dimension. Abbreviations - Under N_Ff: QF = quadriceps femoris, BF = biceps femoris, S = soleus, LD = latissimus dorsi, a = activation, s = synchronization; Under O_Ff: S = saccades, Fix = fixations, v = velocity, n = number, m = mean, sd = standard deviation, GA = gaze area; Under B_Ff: A = ankle, K = knee, H = hip, T = Trunk, SAG = sagittal plane, FR = frontal plane, TRV = transverse plane; Under E_Ff: EX = exhertion, PF = performance, FR = frustration, per = perceived; Under L_Ff: TtT = time to target, COL = collisions, SPD = speed, IM = impulse, TR = turning rate, DT = distance to NPC, AC = acceleration, TN = turns, m = mean, mx = maximum, mn = minimum, sd = standard deviation, n = number.

Figure 9.

A sample 3’ × 4’ ChArUco board (each square equates to 1 ft.2).