|
Principal Investigator:
|
Emery N. Brown, M.D., Ph.D.
|
|
Organization:
|
Harvard - Massachusetts General Hospital
|
|
Project Ended:
|
2000
|
Background The goal of this project is to develop reliable statistical algorithms for on-line analysis of physiologic and neurobehavioral variables monitored during long-duration space missions. Maintenance of physiologic and neurobehavioral homeostasis during long-duration space missions is crucial for ensuring optimal crew performance. If countermeasures are not applied, alterations in homeostasis will occur in nearly all-physiologic systems. During such missions data from most of these systems will be either continually and/or continuously monitored. Therefore, if these data can be analyzed as they are acquired and the status of these systems can be continually assessed, then once alterations are detected, appropriate countermeasures can be applied to correct them.
One of the most important physiologic systems in which to maintain homeostasis during long-duration missions is the circadian system. To detect and treat alterations in circadian physiology during long duration space missions requires development of: 1) a ground-based protocol to assess the status of the circadian system under the light-dark environment in which crews in space will typically work; and 2) appropriate statistical methods to make this assessment. The protocol in Dr. Charles Czeisler's project, Circadian Entrainment, Sleep-Wake Regulation and Neurobehavioral Performance During Extended Duration Space Flight, will study human volunteers under the simulated light-dark environment of long-duration space missions. Therefore, we propose to develop statistical models to characterize in near real time circadian and neurobehavioral physiology under these conditions.
The specific aims of this project are to test the hypotheses that: 1) Dynamic statistical methods based on the Kronauer model of the human circadian system can be developed to estimate circadian phase, period, amplitude from core-temperature data collected under simulated light-dark conditions of long-duration space missions. 2) Analytic formulae and numerical algorithms can be developed to compute the error in the estimates of circadian phase, period and amplitude determined from the data in Specific Aim One. 3) Statistical models can detect reliably in near real-time (daily) significant alternations in the circadian physiology of individual subjects by analyzing the circadian and neurobehavioral data collected in Dr. Czeisler's project. 4) Criteria can be developed using the Kronauer model and the recently developed Jewett model of cognitive performance and subjective alertness to define altered circadian and neurobehavioral physiology and to set conditions for immediate administration of countermeasures.
Research Plan Years Two and Three
At the outset of Year Two we made three changes in the research plan as a consequence of the research findings in Year One and the recommendations of the review committee.
Change 1: Dynamic Assessments of Circadian Phase from Forced Desynchrony Studies. In our Year One research plan, our original goal was to use the data collected during the 25 24-hour days of core-temperature data collected from Dr. Czeisler's project to develop a technique for making dynamic assessments of circadian phase. These estimates would provide the circadian input to the performance and subjective alertness model prediction developed by Dr. Jewett. Our original hypothesis was that under low light conditions these subjects would free run and therefore these data would provide an excellent framework for making dynamic assessments of circadian phase. All of the three subjects analyzed by the end of Year One were entrained during the 25 24-hour day. Our analysis and the independent constant routine assessments confirmed this. Therefore to test the ability of our analytic framework to make dynamic assessments of circadian phase we use the temperature data from the forced desynchrony part of the protocol. During this phase of the protocol the subject is desynchronized from the 28-hour day.
Change 2: Average Prediction of Performance and Subjective Alertness. In our Year One research plan our original goal was to develop straight away an algorithm for making time specific individual predictions of performance and subjective alertness using the models developed by Dr. Jewett. We realized that moving directly to individual predictions was too large an initial step. Therefore we will use the performance, alertness and circadian phase data to first adapt the Jewett model to predict average performance, since this is what it was initially developed to predict. Once the model shows good predictions with average performance and subjective alertness, we will then return the problem of individual predictions.
Change 3: Use the Expertise of a Neurobehavioralist. Our scientific review committee recommended that we include a neurobehavioralist on our team in order to better focus the work on performance and subjective alertness. In response to this suggestion, we have Dr. Megan Jewett working on this component of the modeling for the project. She developed the performance and subjective alertness models for her doctoral dissertation and has been working with us to adapt them to the study of the subjects on the simulated long-duration space missions.
The objective in Year Three was to analyze the core-temperature, performance and alertness data of the seven subjects in the control group from Dr. Czeisler's project.
Progress, Results and Implications for Future Research
Core-Temperature Analysis. The methods were applied to the seven control subjects studied in Dr. Czeisler's project. We have successfully used our methods to analyze core-temperature data on the forced desynchrony protocol and demonstrate that the period of the human circadian pacemaker is closer to 24 instead of 25 hours. We published these findings in Science in July of 1999. In addition we published two manuscripts detailing our methods for dynamic assessment of circadian phase.
Genetic Algorithm. We made the genetic algorithm a standard part of our analysis framework. It has been implemented with a continuous state discrete-time Kalman filter algorithm in order to fit unevenly spaced core-temperature data.
New Model for Core-Temperature. The new core-temperature model described in our two publications holds promise for giving a better description of the dynamics of the human circadian pacemaker. This description can be further enhanced if realistic models of the thermoregulatory and activity interactions with the circadian pacemaker can be characterized. We will work on developing these model components.
Analysis of Performance and Subjective Alertness. We are completing our analysis of subjective alertness and performance for all of the subjects in the control group in Dr. Czeilser's project. A manuscript on this work in under preparation.
Growth Hormone Model. We have developed a growth hormone model so that we can now include the negative feedback to measure the effect of growth hormone plasma levels on its own secretion. We are using the model to analyze normal growth hormone physiology in normal subjects.
Cortisol Model. We have also submitted for publication a manuscript detailing a new stochastic differential equation model of plasma cortisol levels. This model may be used to analyze diurnal cortisol patterns as well as serving as a starting point for extending our work on growth hormone to the analysis of melatonin series with more than one secretory event.
Implications for Future Research. We have developed more accurate statistical models of human circadian marker rhythms including core-temperature, growth hormone and cortisol. These new models provide an accurate means of assessing characterizing circadian physiology with respect to standard marker rhythms in both space and non-space related research. We have used the dynamic phase information from the core-temperature model as an input to the Jewett performance and subjective alertness models. The paradigm may be useful for developing accurate strategies to monitor the circadian health, neurobehavioral state of astronauts during long-duration space missions and for implementing and measuring the effects of countermeasures when significant alterations in these states are detected.
Print