The response time of a biosensor is a crucial metric in safety-critical applications such as medical diagnostics where an earlier diagnosis can markedly improve patient outcomes. However, the speed at which a biosensor reaches a final equilibrium state can be limited by poor mass transport and long molecular diffusion times that increase the time it takes target molecules to reach the active sensing region of a biosensor. While optimization of system and sensor design can promote molecules reaching the sensing element faster, a simpler and complementary approach for response time reduction that is widely applicable across all sensor platforms is to use time-series forecasting to predict the ultimate steady-state sensor response. In this work, we show that ensembles of long short-term memory (LSTM) networks can accurately predict equilibrium biosensor response from a small quantity of initial time-dependent biosensor measurements, allowing for significant reduction in response time by a mean and median factor of improvement of 18.6 and 5.1 respectively. The ensemble of models simultaneously estimates uncertainty, which is vital for ensuring confidence in the predictions and subsequent safety-related decisions that are made. This approach is demonstrated on real-time experimental data collected by exposing porous silicon biosensors to buffered protein solutions using a multi-channel fluidic cell that enables the automated measurement of 100 porous silicon biosensors in parallel. The dramatic improvement in sensor response time achieved using LSTM network ensembles and associated uncertainty quantification opens the door to trustworthy and faster responding biosensors, enabling more rapid medical diagnostics for faster clinical decision making that can lead to improved patient outcomes and healthcare access, as well as quicker identification of toxins in food and the environment.

Porous silicon (PSi) thin films on silicon substrates have been extensively investigated in the context of biosensing applications, particularly for achieving label-free optical detection of a wide range of analytes. However, mass transport challenges have made it difficult for these biosensors to achieve rapid response times and low detection limits. In this work, we introduce an approach for improving the efficiency of molecule transport in PSi by using open-ended PSi membranes atop paper substrates in a flow-through sensor scheme. The paper substrate provides structural support as well as an efficient means of draining solutions from the PSi membrane without the use of an external pump and microfluidic channels. Distinct changes in the reflectance properties of the PSi membrane are measured when molecules are captured in the membrane. A concentration-dependent response of the sensor for protein detection is demonstrated. Factors influencing the interaction time of molecules in the PSi membrane and the drying time of the membrane, which directly affect the detection sensitivity and overall testing time, are discussed. The demonstrated performance of the PSi-on-paper sensor establishes the feasibility of a platform for low-cost rapid diagnostic tests with a highly sensitive, quantitative readout.

We report a generalizable computational approach to dramatically reduce biomolecular and chemicalsensor response time for applications including medical diagnostics. Comparing the performance of different models, we use experimental data to train ensembles of both traditional recurrent neural networks (RNN) and long short-term memory (LSTM) networks, to accurately predict equilibrium sensor response from data measured over a short time span. This approach is particularly advantageous for sensor platforms with long response times due to poor mass transport, including porous silicon optical biosensors, which we use to validate this methodology through exposure to various concentrations of protein solution and subsequent analysis.

We report a versatile platform based on an array of porous silicon (PSi) thin films that can identify analytes based on their physical and chemical properties without the use of specific capture agents. The ability of this system to reproducibly classify, quantify, and discriminate three proteins separately is demonstrated by probing the reflectance of PSi array elements with a unique combination of pore size and buffer pH, and by analyzing the optical signals using machine learning. Protein identification and discrimination are reported over a concentration range of two orders of magnitude. This work represents a significant first step towards a low-cost, simple, versatile, and robust sensor platform that is able to detect biomolecules without the added expense and limitations of using capture agents.

We report the incorporation of porous silicon on paper towards the realization of a low-cost rapid diagnostic testing platform with the capability for quantification of detected molecules. Real-time optical reflectance measurements of bovine serum albumin adsorption were carried out to benchmark the sensor. Simulations and experiments demonstrate a response-time dependence on porous silicon pore size. Approaches to overcome challenges with porous silicon adhesion on paper and porous silicon membrane mechanical robustness will be discussed.

We investigate the utility of various statistical and machine learning techniques for classifying and quantifying selected proteins using an array of porous silicon sensors with uniquely tuned properties. No capture agents or bioreceptors are utilized for the protein detection. The sensing approach relies on differences in non-specific physisorption and represents a step towards a new low cost, simple and robust sensor platform that can detect a vast range of biomolecules.

The ultimate detection limit of optical biosensors is often limited by various noise sources, including those introduced by the optical measurement setup. While sophisticated modifications to instrumentation may reduce noise, a simpler approach that can benefit all sensor platforms is the application of signal processing to minimize the deleterious effects of noise. In this work, we show that applying complex Morlet wavelet convolution to Fabry-Pérot interference fringes characteristic of thin film reflectometric biosensors effectively filters out white noise and low-frequency reflectance variations. Subsequent calculation of the average difference in extracted phase between the filtered analyte and reference signals enables a significant reduction in the limit of detection (LOD). This method is applied on experimental data sets of thin film porous silicon sensors (PSi) in buffered solution and complex media obtained from two different laboratories. The demonstrated improvement in the LOD achieved using wavelet convolution and average phase difference paves the way for PSi optical biosensors to operate with clinically relevant detection limits for medical diagnostics, environmental monitoring, and food safety.

We present a signal processing method capable of significantly lowering the detection limit of thin film optical biosensors. This signal processing method, which we term LAMP, is based on Morlet wavelet filtering and extracting phase information. LAMP drastically reduces noise contributions typically encountered in sensor measurements. The noise immunity, sensitivity and linearity of LAMP is benchmarked against several other signal processing methods by applying the different techniques to a large set of simulated porous silicon thin film optical spectra that contain various types of noise signatures. The LAMP signal processing technique opens up new applications in disease detection and environmental monitoring for thin film sensors previously precluded by insufficient detection limits.

The anterior gradient homologue-2 (AGR2) protein is an attractive biomarker for various types of cancer. In pancreatic cancer, it is secreted to the pancreatic juice by premalignant lesions, which would be an ideal stage for diagnosis. Thus, designing assays for the sensitive detection of AGR2 would be highly valuable for the potential early diagnosis of pancreatic and other types of cancer. Herein, we present a biosensor for label-free AGR2 detection and investigate approaches for enhancing the aptasensor sensitivity by accelerating the target mass transfer rate and reducing the system noise. The biosensor is based on a nanostructured porous silicon thin film that is decorated with anti-AGR2 aptamers, where real-time monitoring of the reflectance changes enables the detection and quantification of AGR2, as well as the study of the diffusion and target-aptamer binding kinetics. The aptasensor is highly selective for AGR2 and can detect the protein in simulated pancreatic juice, where its concentration is outnumbered by orders of magnitude by numerous proteins. The aptasensor's analytical performance is characterized with a linear detection range of 0.05-2 mg mL-1, an apparent dissociation constant of 21 ± 1 μM, and a limit of detection of 9.2 μg mL-1 (0.2 μM), which is attributed to mass transfer limitations. To improve the latter, we applied different strategies to increase the diffusion flux to and within the nanostructure, such as the application of isotachophoresis for the preconcentration of AGR2 on the aptasensor, mixing, or integration with microchannels. By combining these approaches with a new signal processing technique that employs Morlet wavelet filtering and phase analysis, we achieve a limit of detection of 15 nM without compromising the biosensor's selectivity and specificity.