RTD Late is the rate of force development across the 100 to 200 ms window after onset. The platform computes it the same way it computes RTD Early: a two-point slope with both endpoints linearly interpolated to land exactly on the target times, no curve fit, no goodness-of-fit statistic. The math is straightforward. The clinical reason RTD Late exists alongside RTD Early is that the two windows reflect different physiology, and reading them together unpacks a contraction in ways neither one can do alone.
If you haven't already, what RTD Early actually means covers the algorithm and the sample-rate sensitivity that applies equally here. This article focuses on what makes the late window distinct.
The definition
RTD Late is the slope between the interpolated force at onset + 100 ms and the interpolated force at onset + 200 ms. Units are Nm/s when torque conversion is available, N/s otherwise.
rtdLate = (force_at_200ms − force_at_100ms) / 0.100s
The 100 ms window length is twice as long as the 75 ms early window, which gives the late tile slightly easier sample-density requirements. At 40 Hz the late window contains 4 samples (moderate reliability) where the early window contains 3 (low). At 80 Hz the late window contains 8 samples (high), where the early window has 6 (also high). The reliability badge on the late tile uses the same tiering logic as the early tile but applied to the 100 ms window.
Why the late window reflects different physiology
The neuromuscular system that produces a maximum-effort contraction is layered. In the first 50 to 75 ms, motor unit recruitment speed and the maximal discharge rate of the motor neuron pool dominate the force trajectory; the muscle's own contractile machinery has barely begun to catch up. By 100 to 200 ms, the picture has inverted. Motor unit discharge rates have plateaued at similar levels across individuals, and what differentiates patients' force production is the muscle's intrinsic contractile properties.
The published evidence on this is concrete. Decomposing burst-like explosive contractions of the knee extensors:
- Force at 50 ms is explained almost entirely by EMG amplitude in the first 50 ms (R² = 0.36, neural drive).
- Force at 100 ms is explained primarily by the electrically evoked octet response (R² = 0.65, pure contractile).
- Force at 150 ms is explained primarily by maximum voluntary force itself (R² = 0.71), with a small additional octet contribution (R² = 0.06).
Source: D'Emanuele et al. 2023. The early-late distinction in RFD windows is not a stylistic convention; it's a clean dissociation of underlying mechanisms, and the platform surfaces both because the published evidence dissociates both.
Reading early and late together
Three patterns recur enough to be worth naming:
Concordant high. Both RTD Early and RTD Late are at or near the patient's healthy baseline (or the uninvolved side). The neural and contractile components are tracking together. This is what recovery looks like.
Concordant low. Both windows are suppressed relative to baseline or the uninvolved side. The deficit is global; the patient is producing force slowly in both phases. Often early in rehab, often when peak strength is also down. Treatment focuses on building capacity across the board.
Discordant. One window is recovered, the other is not. The two flavors:
- RTD Early recovered, RTD Late suppressed. The patient can fire fast, but the contractile machinery is not keeping pace. The literature on post-ACL knee extension shows this pattern: standardized mean differences for early-phase RFD versus the uninvolved limb run around −1.07, and late-phase SMDs run around −0.85, both suppressed but with the early window often returning toward baseline first (Turpeinen et al. 2020). Recovery of RTD Late lags recovery of RTD Early.
- RTD Early suppressed, RTD Late recovered. Less common. Suggests the neural drive phase is the limiting factor while the muscle's intrinsic contractile properties are intact. Sometimes seen with arthrogenic inhibition or persistent quadriceps activation deficits.
The radar on the LSI panel makes these patterns visible at a glance: the early and late tiles sit next to each other on the radar shape, and discordance between them shows up as a step in the radar's outline.
The corticomotor connection
There's also a neural-imaging thread that aligns with the early-late mechanistic split. In individuals after ACL surgery, the active motor threshold (a transcranial-magnetic-stimulation measure of corticospinal excitability) correlates with RTD across 0 to 50 ms (r = −0.51), while short-interval intracortical inhibition correlates with RTD across 100 to 200 ms (r = 0.50) (Scheurer et al. 2020). The two windows have distinct corticomotor correlates: corticospinal pathways predict the early window, intracortical inhibition predicts the late window. The platform doesn't measure either, but it's a useful frame for understanding why the early-late dissociation reads as more than a measurement convention.
When the relationship matters most
Three clinical situations where the early-late pair is more informative than either alone:
- Return-to-sport gating. A patient at 95% peak LSI and 70% RTD Early LSI is a different clinical picture than a patient at 95% peak LSI and 70% RTD Late LSI. The first has a neural drive deficit; the second has a contractile deficit. Both need RFD work, but the dosing and modality should differ. The LSI article covers how the platform surfaces the discordance.
- Tracking adaptations to a training block. Early adaptations to explosive strength training often show up first in RTD Late (the contractile phase responds faster to volume) before propagating to RTD Early. Watching both windows over a block of training lets you see whether the patient is adapting in the right place.
- Disuse recovery. Periods of unloading or disuse suppress RTD disproportionately faster than maximum strength (Ruggiero & Gruber 2024), and the late window often suppresses more than the early window over short unloading periods. Tracking both during the return-to-load phase makes the recovery trajectory legible.
What to do next
- Read RTD Late alongside RTD Early on every session, not in isolation. The relationship is the diagnostic.
- For low-rate devices, watch the reliability badge on both tiles. The 100 ms window is slightly more forgiving than the 75 ms window, but both can drop into low confidence at 40 Hz.
- A green peak with discordant RTD windows is a real pattern. Surface it in the chart note; don't let the headline strength number obscure it.
References
- D'Emanuele S, Tarperi C, Rainoldi A, et al. Neural and contractile determinants of burst-like explosive isometric contractions of the knee extensors. Scand J Med Sci Sports. 2023;33(2):127-135. doi:10.1111/sms.14244
- Del Vecchio A. Neuromechanics of the rate of force development. Exerc Sport Sci Rev. 2022. doi:10.1249/JES.0000000000000306
- Turpeinen JT, Freitas TT, Rubio-Arias JÁ, et al. Contractile rate of force development after anterior cruciate ligament reconstruction, a comprehensive review and meta-analysis. Scand J Med Sci Sports. 2020;30(9):1572-1585. doi:10.1111/sms.13733
- Scheurer SA, Sherman DA, Glaviano NR, et al. Corticomotor function is associated with quadriceps rate of torque development in individuals with ACL surgery. Exp Brain Res. 2020;238(2):283-294. doi:10.1007/s00221-019-05713-w
- Ruggiero L, Gruber M. Neuromuscular mechanisms for the fast decline in rate of force development with muscle disuse, a narrative review. J Physiol. 2024. doi:10.1113/JP285667