Investigating Risk Factors and Predicting Complications in Deep Brain Stimulation Surgery with Machine Learning Algorithms.

BACKGROUND: Deep brain stimulation (DBS) surgery is an option for patients experiencing medically resistant neurologic symptoms. DBS complications are rare; finding significant predictors requires a large number of surgeries. Machine learning algorithms may be used to effectively predict these outcomes. The aims of this study were to 1) investigate preoperative clinical risk factors and 2) build machine learning models to predict adverse outcomes. METHODS: This multicenter registry collected clinical and demographic characteristics of patients undergoing DBS surgery (n = 501) and tabulated occurrence of complications. Logistic regression was used to evaluate risk factors. Supervised learning algorithms were trained and validated on 70% and 30%, respectively, of both oversampled and original registry data. Performance was evaluated using area under the receiver operating characteristics curve (AUC), sensitivity, specificity, and accuracy. RESULTS: Logistic regression showed that the risk of complication was related to the operating institution in which the surgery was performed (odds ratio [OR] = 0.44, confidence interval [CI] = 0.25-0.78), body mass index (OR = 0.94, CI = 0.89-0.99), and diabetes (OR = 2.33, CI = 1.18-4.60). Patients with diabetes were almost 3× more likely to return to the operating room (OR = 2.78, CI = 1.31-5.88). Patients with a history of smoking were 4× more likely to experience postoperative infection (OR = 4.20, CI = 1.21-14.61). Supervised learning algorithms demonstrated high discrimination performance when predicting any complication (AUC = 0.86), a complication within 12 months (AUC = 0.91), return to the operating room (AUC = 0.88), and infection (AUC = 0.97). Age, body mass index, procedure side, gender, and a diagnosis of Parkinson disease were influential features. CONCLUSIONS: Multiple significant complication risk factors were identified, and supervised learning algorithms effectively predicted adverse outcomes in DBS surgery.

Dysgeusia in deep brain stimulation for essential tremor.

Dysgeusia, or foul taste, is a rarely reported side effect in patients who have undergone deep brain stimulation (DBS) in the thalamus for essential tremor. This retrospective study evaluated the incidence, nature, neurophysiological and anatomical location of dysgeusia following DBS. Of 52 patients who had undergone DBS for essential tremor, eight (15%) reported dysgeusia, which was described as a "metallic," "sour," "foul," or "cold" taste in the mouth. Dysgeusia was separate and distinguishable from paresthesia. Dysgeusia was more frequently reported with bilateral than unilateral DBS implants (6 of 27 (22%) vs. 2 of 25 (8%) patients, respectively). The anatomical locations of the contacts causing dysgeusia were measured on postoperative MRI, and compared to those from seven control patients who did not experience dysgeusia after receiving bilateral DBS implants. Leads causing dysgeusia were more posterior than non-dysgeusia-associated leads (4.5 ±â€¯1.2 vs. 5.7 ±â€¯1.8 mm anterior to the posterior commissure, respectively, P < .001). Intraoperative microelectrode recording indicated that these contacts were in the sensory region of the thalamus. Intraoperative testing found that low sensory threshold for paresthesia predicted the development of dysgeusia postoperatively (1.5 ±â€¯0.5 V vs. 3.3 ±â€¯1.9 V; P < .05). These data indicate that taste perception can be altered in the human through DBS, with posterior leads likely within the sensory region of the thalamus. Dysgeusia can be reduced by changing stimulation parameters, or surgical revision of the lead.

Stereotactic Accuracy and Surgical Utility of the O-Arm in Deep Brain Stimulation Surgery.

BACKGROUND: The stereotactic accuracy of intraoperative imaging is critical to clinical outcome, particularly in "asleep" deep brain stimulation (DBS) surgery that typically forgoes neurophysiological techniques. Different intraoperative imaging modalities and associated accuracies have been reported, including magnetic resonance imaging (MRI), computed tomography (CT), and O-arm. OBJECTIVE: To analyze intraoperative O-arm imaging accuracy and to evaluate the utility of microelectrode mapping. METHODS: O-arm images of DBS electrodes were collected during implantation in the subthalamic nucleus in patients with Parkinson disease. Images were fused to postoperative MRI and postoperative CT scans. Stereotactic coordinates for the electrode tip were measured independently. Radial distances between the images were compared. The impact of microelectrode mapping on final DBS electrode positioning was also evaluated. RESULTS: In 71 consecutive DBS electrodes, the average radial error of the electrode tip between the O-arm and MRI was 1.55 ± 0.58 mm. The average radial error between the O-arm and CT was 1.03 ± 0.61 mm. Thus, the O-arm images accurately depicted the position of the electrode. However, in 14% of cases, microelectrode mapping revised the DBS electrode position beyond the preoperative direct target in combination with accurate intraoperative imaging. CONCLUSION: Intraoperative O-arm images reliably and accurately displayed the location of the DBS electrode compared with postoperative CT and MRI images. Microelectrode mapping provided superior subnuclear resolution to imaging. Both intraoperative imaging and microelectrode mapping are effective tools that can be synergistically combined for optimal DBS electrode placement.

Measuring and modeling reduction of DDT availability to the water column and mussels following activated carbon amendment of contaminated sediment.

A 28-day accumulation study demonstrated the use of mussel uptake, passive samplers, and biodynamic modeling to measure the reduction of dichlorodiphenyltrichloroethane (DDT) availability in the water column after the addition of activated carbon to contaminated sediment. Sediment collected from Lauritzen Channel, Richmond, California (16.5mg total DDT/kg) was mixed with either virgin activated carbon or a reactivated carbon for one month, after which a 28-day laboratory exposure study was completed. Mussels (Mytilus edulis) suspended above activated carbon-treated sediment accumulated significantly less total DDT in soft tissue, 91% and 84% for virgin and reactivated carbon, respectively, as compared to untreated sediment. Mussel tissue concentrations correlated to concentrations in semipermeable membrane devices (SPMDs) and polyethylene devices (PEDs) suspended over the same sediments. A biodynamic model that incorporated DDT water concentrations, either analytically measured or estimated from PED uptake, described mussel accumulation over time. Thus, passive samplers in combination with biodynamic modeling may provide an important screening tool for assessment of filter-feeding uptake and ecological risk to water-dwelling organisms exposed to aqueous phase hydrophobic organic contaminants.

Biodynamic modeling of PCB uptake by Macoma balthica and Corbicula fluminea from sediment amended with activated carbon.

Activated carbon amendment was assessed in the laboratory as a remediation strategy for freshwater sediment contaminated with polychlorinated biphenyls (PCBs) from the Grasse River (near Massena, NY). Three end points were evaluated: aqueous equilibrium PCB concentration, uptake into semipermeable membrane devices (SPMDs), and 28-day bioaccumulation in the clam Corbicula fluminea. PCB uptake by water, SPMDs, and clams followed similar trends, with reductions increasing as a function of carbon dose. Average percent reductions in clam tissue PCBs were 67, 86, and 95% for activated carbon doses of 0.7, 1.3, and 2.5% dry wt, respectively. A biodynamic model that incorporates sediment geochemistry and dietary and aqueous uptake routes was found to agree well with observed uptake by C. fluminea in our laboratory test systems. Results from this study were compared to 28-day bioaccumulation experiments involving PCB-contaminated sediment from Hunters Point Naval Shipyard (San Francisco Bay, CA) and the clam Macoma balthica. Due to differences in feeding strategy, M. balthica deposit-feeds whereas C. fluminea filter-feeds, the relative importance of the aqueous uptake route is predicted to be much higher for C. fluminea than for M. balthica. Whereas M. balthica takes up approximately 90% of its body burden through sediment ingestion, C. fluminea only accumulates approximately 45% via this route. In both cases, results strongly suggest that it is the mass transfer of PCBs from native sediment to added carbon particles, not merely reductions in aqueous PCB concentrations, that effectively reduces PCB bioavailability and uptake by sediment-dwelling organisms.

Bioaccumulation of perfluorochemicals in sediments by the aquatic oligochaete Lumbriculus variegatus.

Bioaccumulation of perfluoroalkyl sulfonates, perfluorocarboxylates, and 2-(N-ethylperfluorooctane sulfonamido) acetic acid (N-EtFOSAA) from laboratory-spiked and contaminated field sediments was assessed using the freshwater oligochaete, Lumbriculus variegatus. Semistatic batch experiments were conducted to monitor the biological uptake of these perfluorochemicals (PFCs) over 56 days. The elimination of PFCs was measured as the loss of PFCs in L. variegatus exposed to PFC-spiked sediment for 28 days and then transferred to clean sediment. The resultant data suggest that PFCs in sediments are readily bioavailable and that bioaccumulation from sediments does not continually increase with increasing perfluorocarbon chain length. Perfluorooctane sulfonate (PFOS) and perfluorononanoate were the most bioaccumulative PFCs, as measured by laboratory-based estimated steady-state biota sediment accumulation factors (BSAFs) and BSAFs measured using contaminated field sediments. Elimination rate constants for perfluoroalkyl sulfonates and perfluorocaroboxylates were generally smaller than those previously measured for other organic contaminants. Last, a PFOS precursor, N-EtFOSAA, accumulated in the worm tissues and appeared to undergo biotransformation to PFOS and other PFOS precursors. This suggests that N-EtFOSAA, which has been detected in sediments and sludge often at levels exceeding PFOS, may contribute to the bioaccumulation of PFOS in aquatic organisms.

Biological uptake of polychlorinated biphenyls by Macoma balthica from sediment amended with activated carbon.

This work characterizes the efficacy of activated carbon amendment in reducing polychlorinated biphenyl (PCB) bioavailability to clams (Macoma balthica) from field-contaminated sediment (Hunters Point Naval Shipyard, San Francisco Bay, CA, USA). Test methods were developed for the use of clams to investigate the effects of sediment amendment on biological uptake. Sediment was mixed with activated carbon for one month. Bioaccumulation tests (28 d) were employed to assess the relationships between carbon dose and carbon particle size on observed reductions in clam biological uptake of PCBs. Extraction and cleanup protocols were developed for the clam tissue. Efficacy of activated carbon treatment was found to increase with both increasing carbon dose and decreasing carbon particle size. Average reductions in bioaccumulation of 22, 64, and 84% relative to untreated Hunters Point sediment were observed for carbon amendments of 0.34, 1.7, and 3.4%, respectively. Average bioaccumulation reductions of 41, 73, and 89% were observed for amendments (dose = 1.7% dry wt) with carbon particles of 180 to 250, 75 to 180, and 25 to 75 microm, respectively, in diameter, indicating kinetic phenomena in these tests. Additionally, a biodynamic model quantifying clam PCB uptake from water and sediment as well as loss through elimination provided a good fit of experimental data. Model predictions suggest that the sediment ingestion route contributed 80 to 95% of the PCB burdens in the clams.

Effects of particulate carbonaceous matter on the bioavailability of benzo[a]pyrene and 2,2',5,5'-tetrachlorobiphenyl to the clam, Macoma balthica.

We investigated the bioavailability via diet of spiked benzo[a]pyrene (BaP) and 2,2',5,5'-tetrachlorobiphenyl (PCB-52) from different carbonaceous (non-carbonate, carbon containing) particle types to clams (Macoma balthica) collected from San Francisco Bay. Our results reveal significant differences in absorption efficiency between compounds and among carbonaceous particle types. Absorption efficiency for PCB-52 was always greater than that for BaP bound to a given particle type. Among particles, absorption efficiency was highest from wood and diatoms and lowest from activated carbon. Large differences in absorption efficiency could not be simply explained by comparatively small differences in the particles' total organic carbon content. BaP and PCB-52 bound to activated carbon exhibited less than 2% absorption efficiency and were up to 60 times less available to clams than the same contaminants associated with other types of carbonaceous matter. These results suggest that variations in the amount and type of sediment particulate carbonaceous matter, whether naturally occurring or added as an amendment, will have a strong influence on the bioavailability of hydrophobic organic contaminants. This has important implications for environmental risk assessment, sediment management, and development of novel remediation techniques.