About

Christopher Rose is a Professor of Engineering at Brown University and serves as the Director of STEMJazz Programs. His research interests encompass communication theory as an organizing principle, wireless communication, molecular communication, mobile networks, communication theory and physics, genetic algorithms, interference avoidance, and MIMO technology. Rose has a broad view of his field, considering it on a cosmic scale, and is involved in advancing understanding and innovation within these areas. He has been recognized for his innovative teaching and support for students, receiving the IEEE Undergraduate Teaching Award, and has been honored with university awards for student advising and mentoring. In addition to his research and teaching, Rose has held administrative roles, including being named Associate Provost at Brown University. His work and contributions are noted in various news articles and university communications, highlighting his influence in both academic and practical aspects of communication technology.

Research topics

• Computer Science
• Database
• Chemistry
• Combinatorial chemistry

Selected publications

• Limits of Information Flow Between Classically Interacting Particles

  PubMed · 2025-01-30

  preprintOpen accessSenior author

  Pinning down a precise understanding of information flow within physical interactions remains a central concern to fields such as stochastic thermodynamics and quantum information science. In both spheres a careful accounting of bits (or qubits) enables a deeper understanding of the physical nature of information. In this work we propose a measure of information flow as a saddle-point solution of the mutual information. This approach places a lower bound on the channel capacity between a particle and an interacting environment. The measure is given by P_{0}/2E_{0} in nats/sec, with P_{0} the average power flux between the particle and its environment, and E_{0} the initial average energy of the particle, all computed in a frame where the particle has zero average momentum. We use a communication theory lens to suggest an associated channel analogy, in which this bound is interpreted as a signal-to-noise ratio. We find that this measure can also quantify early-time information flow for a particle interacting with a thermal bath.

  PublisherOA PDFDOI

• Using a large language model (ChatGPT) to assess risk of bias in randomized controlled trials of medical interventions: protocol for a pilot study of interrater agreement with human reviewers

  BMC Medical Research Methodology · 2025-07-31 · 2 citations

  articleOpen access1st authorCorresponding

  BACKGROUND: Risk of bias (RoB) assessment is an essential part of systematic reviews that requires reading and understanding each eligible trial and RoB tools. RoB assessment is subject to human error and is time-consuming. Machine learning-based tools have been developed to automate RoB assessment using simple models trained on limited corpuses. ChatGPT is a conversational agent based on a large language model (LLM) that was trained on an internet-scale corpus and has demonstrated human-like abilities in multiple areas including healthcare. LLMs might be able to support systematic reviewing tasks such as assessing RoB. We aim to assess interrater agreement in overall (rather than domain-level) RoB assessment between human reviewers and ChatGPT, in randomized controlled trials of interventions within medical interventions. METHODS: We will randomly select 100 individually- or cluster-randomized, parallel, two-arm trials of medical interventions from recent Cochrane systematic reviews that have been assessed using the RoB1 or RoB2 family of tools. We will exclude reviews and trials that were performed under emergency conditions (e.g., COVID-19), as well as public health and welfare interventions. We will use 25 of the trials and human RoB assessments to engineer a ChatGPT prompt for assessing overall RoB, based on trial methods text. We will obtain ChatGPT assessments of RoB for the remaining 75 trials and human assessments. We will then estimate interrater agreement using Cohen's κ. RESULTS: The primary outcome for this study is overall human-ChatGPT interrater agreement. We will report observed agreement with an exact 95% confidence interval, expected agreement under random assessment, Cohen's κ, and a p-value testing the null hypothesis of no difference in agreement. Several other analyses are also planned. CONCLUSIONS: This study is likely to provide the first evidence on interrater agreement between human RoB assessments and those provided by LLMs and will inform subsequent research in this area.

  PublisherOA PDFDOI

• A Novel Instrument Design for Studying Photometric Effects in Reflectance and Emission Data in Visible and Mid-IR Wavelengths

  2025-07-09

  preprintOpen accessSenior author

  Background and MotivationDecades of space missions have advanced our understanding of planetary surfaces through remote sensing and in-situ spectroscopy [1, 2, 3]. These efforts produce high-quality data but suffer from observational limitations and often requiring complex models and lab validation [4, 5, 6]. A major source of uncertainty is the effect of viewing geometry on spectral measurements, causing modifications in spectral features such as shape, depth, and slope, among others [7, 8, 9, 10]. This can cause various issues with misidentification of surface materials [11, 12]. Phase angle variations influence surface brightness and albedo, complicating surface temperature measurements from instruments like Diviner on the Lunar Reconnaissance Orbiter [10]. These complications have also been seen on Mars as phase reddening and variable brightness temperatures [13, 14]. Here, we present our Spectrogoniometer designed to measure effects of varying phase angles on spectral measurements and the ”first-light” data.The instrument presented here supports both reflectance and emission measurements. In reflectance mode, the viewing geometry is defined by the phase angle (ϕ, Figure 1), which depends on the incidence angle (i, Figure 1), emergence angle (α, Figure 1), and their respective azimuthal directions (θ, Figure 1) [4]. In emission, it is characterized by the emission angle (e, Figure 1) [4]. Laboratory studies of viewing angle effects are limited due to challenges in instrumentation and automation. Most lab measure ments use fixed geometries (e.g., 30◦ phase angle with no azimuthal offset), while mission data has often relied heavily on modeled data for interpretation [15, 16]. Efforts to characterize these effects have focused primarily on the visible (VIS) and short-waveIR (SWIR) ranges, with growing work in longer IR wavelengths (e.g. Figure 1) [17, 18]. Additionally, most existing setups only utilize reflectance, lacking emission capabilities [e.g. 18].To support spacecraft data, lab measurements must replicate similar phase angles. Rovers and landers, unlike orbiters, frequently view surfaces at high emergence/emission angles. These large phase angles complicate comparison to nadir-viewing instruments and hinder data interpretation by affecting the angular distribution of scattered light, which depends on grain size, shape, and internal structure [5, 6]. Absorbing surface layers and small geometric changes at high angles further amplify scattering complexities and cause significant variations in the phase curve, making accurate analysis more difficult [6].Due to the lack of well-characterized spectral measurements across varying phase angles, high-phase-angle data is often interpreted through complex models that rely heavily on estimations and assumptions. While a dedicated phase angle measurement campaign won’t eliminate these complexities, it can provide valuable constraints and improve model accuracy, enhancing space craft data interpretation.With the growing volume of spacecraft observations and measurements, including past, current, and future missions, the need to collect lab data that matches observation geometries is increasingly important. Figure 2 ([19]) illustrates limited laboratory measurements across a small phase angle range, but it highlights an example of phase angle characterization in a laboratory setting. Even in the case of a few phase angles measured, there are spectral features that are changing or in some cases disappearing from the spectra.Spectrogoniometers capable of measuring variable phase angles do exist [e.g. 17, 18], but [18] is restricted to VIS and SWIR ranges. Though helpful, current instruments have limited capacity to support a broad range of mission science objectives. Specifically, phase angle measurements in the thermal IR (TIR) remain largely unsupported in a laboratory setting. The instrument described in [17] was originally designed to cover both VIS and IR, but [20] notes theIR capability is no longer functional. Improved characterization of phase angle effects across a wider spectral range will enhance the support and accuracy of modeling efforts used in the interpretation of both remote sensing and in-situ data.Methodology and Instrument DescriptionThe spectrogoniometer used in this work (Figure 3) is a multi-wavelength spectrogoniometer with a measurable spectral range of .285 μm - 18 μm with a spectral resolution of approximately 1cm−1. The instrument utilizes a dual arm design. One arm, the receiving arm (Rx Arm) "A", uses a series of relay optics "B" to collect light reflected or emitted from the sample to the detector. The transmit arm (Tx Arm) "C", holds the light source "D" used in reflectance measurements. The motion of each arm is depicted in Figure 4. Both arms have an approximate elevation angle range of ±70◦. The Tx arm has an azimuth range of 230◦. The sample tray "F" contains four fixed positions and rotates to bring each position under the arms. It contains the sample and three calibration targets. This design allows for efficient measurements and immediate calibration as needed for both reflectance and emission modes.Data collection and calibration was performed using an automated system developed for the instrument. The Rx arm moves incrementally over its whole range then the Tx arm increments and the Rx arm repeats its range. After the Tx arm completes its elevation range, it moves in azimuth and the process is repeated until a measurement is performed at each combination of arm elevations and azimuth. In total, 14 incidence angles, 23 incidence azimuths, 14 emergence angles, and 28 emission angles are characterized per material. Each sample measured produces 4508 reflectance spectra and 28 emission spectra at various phase angle combinations. Data are processed using a pipeline developed for the instrument. For each material analyzed, a hyperspectral reflectance spectrum, a hyperspectral emission spectrum, and visible images are produced. Those data products will be used to identifymodifications (among other things) in spectra as phase angles change. Well characterized materials (e.g. olivine, pyroxenes, clays, etc.) were selected for the initial characterization of this instrument. Later, other materials commonly found in spectral databases (e.g. ASU spectral library [21]) will be analyzed to provide a number of phase angle characterizations for these already well characterized materials. The data produced by this instrument will be analyzed and produce science products which will aid in interpretations of spacecraft-based measurements (e.g. identify commonly altered spectral features, identify phase reddening in materials, surface brightness temperatures, etc.).

  PublisherDOI

• Quantum Analogues for Two Simple Classical Channels

  arXiv (Cornell University) · 2024-06-16

  preprintOpen accessSenior author

  We present some of the peculiar dynamics of two simple sans-entanglement quantum communication channels in a digestible form. Specifically, we contrast the classical gaussian additive channel to its quantum analogue and find that the quantum version features a capacity with interesting time dependence and counterintuitive effects of quantization. We also consider a simple two-level system and comment on the time dependence of its capacity.

  PublisherOA PDFDOI

• Evaluating the effectiveness of brief anti-speeding messages disseminated within warning notices to driving offenders

  Transportation Research Interdisciplinary Perspectives · 2024-06-26 · 5 citations

  articleOpen access

  Road safety messaging represents a long-standing strategy in efforts to reduce speeding, a behaviour which accounts for an estimated 30 % of all road crashes on Australian roads. This study aimed to further examine the effectiveness of such a strategy via the use of a novel message medium that included anti-speeding messaging disseminated to speeding offenders via warning notices issued by a Start Government Transport agency. Informed by the Step approach to Message Design and Testing (SatMDT), survey content was devised to compare the effectiveness of the messages across various outcome measures including intentions to stay within the posted speed limit. A total of 219 speeding offenders completed the online survey. Overall, the results found some positive responses regarding Intervention condition participants’ responses to the anti-speeding messaging featured within the warning notices. Among some of the key findings were that the Intervention condition participants reported relatively low levels of message rejection. Additionally, females responded more favourably to both messages than males. Collectively, the findings suggest that there is no harm in including such messages within warning notices providing such messages are concept-tested prior to use. Overall, the findings suggest this approach may represent a low-cost option for targeted message delivery.

  PublisherDOI

• Evaluating evolution as a learning algorithm

  arXiv (Cornell University) · 2022-08-01

  preprintOpen access

  We interpret the Moran model of natural selection and drift as an algorithm for learning features of a simplified fitness landscape, specifically genotype superiority. This algorithm's efficiency in extracting these characteristics is evaluated by comparing it to a novel Bayesian learning algorithm developed using information-theoretic tools. This algorithm makes use of a communication channel analogy between an environment and an evolving population. We use the associated channel-rate to determine an informative population-sampling procedure. We find that the algorithm can identify genotype superiority faster than the Moran model but at the cost of larger fluctuations in uncertainty.

  PublisherOA PDFDOI

• Hide and Seek On A Budget

  2022 IEEE International Symposium on Information Theory (ISIT) · 2022-06-26

  article1st authorCorresponding

  Imagine a game of hide and seek with an energy-limited hider and an observation rate-limited seeker. We first show that hider energy expenditure is a function of the process covariance and thus, unsurprisingly, that Gaussian motion processes maximize hider entropy, a proxy for seeker effort. We formulate the optimization in terms of the process position (or velocity) spectral density. If the seeker knows when the hider has looked, the optimal motion strategy is a satisfyingly simple renewal process – after an observation, pick a velocity at random and proceed until the next observation where the process renews. However, if the seeker is oblivious to seeker attentions the problem is more complicated. While the power spectral density of the motion process is necessarily bandlimited owing to the energy constraint, a general closed form eluded us. So, we consider a few special cases for both stationary motion processes and a class of non-stationary motion processes.

  PublisherDOI

• A protocol to evaluate unsupervised text clustering to screen and categorize studies in systematic reviews

  Research Square · 2022-05-12

  preprintOpen accessSenior author
  PublisherOA PDFDOI

• Towards harmonizing clinical linear energy transfer (LET) reporting in proton radiotherapy: a European multi-centric study

  Figshare · 2021-01-01 · 14 citations

  articleOpen access

  Clinical data suggest that the relative biological effectiveness (RBE) in proton therapy (PT) varies with linear energy transfer (LET). However, LET calculations are neither standardized nor available in clinical routine. Here, the status of LET calculations among European PT institutions and their comparability are assessed. Eight European PT institutions used suitable treatment planning systems with their center-specific beam model to create treatment plans in a water phantom covering different field arrangements and fulfilling commonly agreed dose objectives. They employed their locally established LET simulation environments and procedures to determine the corresponding LET distributions. Dose distributions <i>D</i>_1.1 and <i>D_RBE</i> assuming constant and variable RBE, respectively, and LET were compared among the institutions. Inter-center variability was assessed based on dose- and LET-volume-histogram parameters. Treatment plans from six institutions fulfilled all clinical goals and were eligible for common analysis. <i>D</i>_1.1 distributions in the target volume were comparable among PT institutions. However, corresponding LET values varied substantially between institutions for all field arrangements<i>,</i> primarily due to differences in LET averaging technique and considered secondary particle spectra. Consequently, <i>D_RBE</i> using non-harmonized LET calculations increased inter-center dose variations substantially compared to <i>D</i>_1.1 and significantly in mean dose to the target volume of perpendicular and opposing field arrangements (<i>p</i> D_RBEdecreased to that observed for<i> D</i>_1.1<i>.</i> Harmonizing the reported LET among PT centers is feasible and allows for consistent multi-centric analysis and reporting of tumor control and toxicity in view of a variable RBE. It may serve as basis for harmonized variable RBE dose prescription in PT.

  PublisherDOI

• Secret messaging with endogenous chemistry

  Scientific Reports · 2021-07-06 · 2 citations

  articleOpen access

  Data encoded in molecules offers opportunities for secret messaging and extreme information density. Here, we explore how the same chemical and physical dimensions used to encode molecular information can expose molecular messages to detection and manipulation. To address these vulnerabilities, we write data using an object's pre-existing surface chemistry in ways that are indistinguishable from the original substrate. While it is simple to embed chemical information onto common objects (covers) using routine steganographic permutation, chemically embedded covers are found to be resistant to detection by sophisticated analytical tools. Using Turbo codes for efficient digital error correction, we demonstrate recovery of secret keys hidden in the pre-existing chemistry of American one dollar bills. These demonstrations highlight ways to improve security in other molecular domains, and show how the chemical fingerprints of common objects can be harnessed for data storage and communication.

  PublisherOA PDFDOI

Recent grants

• CDI Type I:A Communications Theory Approach to Morphogenesis and Architecture Maintenance

  NSF · $756k · 2008–2012

Frequent coauthors

• Jacob K. Rosenstein

  Providence College

  71 shared

• Eamonn Kennedy

  VA Salt Lake City Healthcare System

  53 shared

• Brenda M. Rubenstein

  Providence College

  44 shared

• Christopher E. Arcadia

  Brown University

  39 shared

• Sherief Reda

  31 shared

• Eunsuk Kim

  Brown University

  28 shared

• Amanda Dombroski

  Brown University

  28 shared

• Pratistha Shakya

  28 shared

Education

• Ph.D., EECS

  Massachusetts Institute of Technology

  1985

• SM, EECS

  Massachusetts Institute of Technology

  1981

• SB, EECS

  Massachusetts Institute of Technology

  1979

Awards & honors

• IEEE Undergraduate Teaching Award (2022)